2020
Ríos-Vila, A.; Calvo-Zaragoza, J.; Iñesta, J. M.
Exploring the two-dimensional nature of music notation for score recognition with end-to-end approaches Proceedings Article
In: Proc. of 17th International Conference on Frontiers in Handwriting Recognition (ICFHR), pp. 193–198, IEEE Computer Society IEEE, Dormund (Germany), 2020, ISBN: 978-1-7281-9966-5.
Abstract | BibTeX | Tags: HispaMus
@inproceedings{k461,
title = {Exploring the two-dimensional nature of music notation for score recognition with end-to-end approaches},
author = {A. Ríos-Vila and J. Calvo-Zaragoza and J. M. Iñesta},
isbn = {978-1-7281-9966-5},
year = {2020},
date = {2020-01-01},
booktitle = {Proc. of 17th International Conference on Frontiers in Handwriting Recognition (ICFHR)},
pages = {193--198},
publisher = {IEEE},
address = {Dormund (Germany)},
organization = {IEEE Computer Society},
abstract = {Optical Music Recognition workflows perform several steps to retrieve the content in music score images, being symbol recognition one of the key stages. State-of-the-art approaches for this stage currently address the coding of the output symbols as if they were plain text characters. However, music symbols have a two-dimensional nature that is ignored in these approaches. In this paper, we explore alternative output representations to perform music symbol recognition with state-of-the-art end-to-end neural technologies. We propose and describe new output representations which take into account the mentioned two-dimensional nature. We seek answers to the question of whether it is possible to obtain better recognition results in both printed and handwritten music scores. In this analysis, we compare the results given using three output encodings and two neural approaches. We found that one of the proposed encodings outperforms the results obtained by the standard one. This permits us to conclude that it is interesting to keep researching on this topic to improve end-to-end music score recognition.},
keywords = {HispaMus},
pubstate = {published},
tppubtype = {inproceedings}
}
Optical Music Recognition workflows perform several steps to retrieve the content in music score images, being symbol recognition one of the key stages. State-of-the-art approaches for this stage currently address the coding of the output symbols as if they were plain text characters. However, music symbols have a two-dimensional nature that is ignored in these approaches. In this paper, we explore alternative output representations to perform music symbol recognition with state-of-the-art end-to-end neural technologies. We propose and describe new output representations which take into account the mentioned two-dimensional nature. We seek answers to the question of whether it is possible to obtain better recognition results in both printed and handwritten music scores. In this analysis, we compare the results given using three output encodings and two neural approaches. We found that one of the proposed encodings outperforms the results obtained by the standard one. This permits us to conclude that it is interesting to keep researching on this topic to improve end-to-end music score recognition. Gallego, A. J.; Calvo-Zaragoza, J.; Rico-Juan, J. R.
Insights Into Efficient k-Nearest Neighbor Classification With Convolutional Neural Codes Journal Article
In: IEEE Access, vol. 8, pp. 99312-99326, 2020, ISSN: 2169-3536.
Abstract | BibTeX | Tags: HispaMus
@article{k444,
title = {Insights Into Efficient k-Nearest Neighbor Classification With Convolutional Neural Codes},
author = {A. J. Gallego and J. Calvo-Zaragoza and J. R. Rico-Juan},
issn = {2169-3536},
year = {2020},
date = {2020-01-01},
urldate = {2020-01-01},
journal = {IEEE Access},
volume = {8},
pages = {99312-99326},
abstract = {The increasing consideration of Convolutional Neural Networks (CNN) has not prevented the use of the k-Nearest Neighbor (kNN) method. In fact, a hybrid CNN-kNN approach is an interesting option in which the network specializes in feature extraction through its activations (Neural Codes), while the kNN has the advantage of performing a retrieval by means of similarity. However, this hybrid approach also has the disadvantages of the kNN search, and especially its high computational cost which is, in principle, undesirable for large-scale data. In this paper, we present the first comprehensive study of efficient kNN search algorithms using this hybrid CNN-kNN approach. This has been done by considering up to 16 different algorithms, each of which is evaluated with a different parametrization, in 7 datasets of heterogeneous composition. Our results show that no single algorithm is capable of covering all aspects, but rather that each family of algorithms is better suited to specific aspects of the problem. This signifies that Fast Similarity Search algorithms maintain their performance, but do not reduce the cost as much as the Data Reduction family does. In turn, the Approximated Similarity Search family is postulated as a good option when attempting to balance accuracy and efficiency. The experiments also suggest that considering statistical transformation algorithms such as Linear Discriminant Analysis might be useful in certain cases.},
keywords = {HispaMus},
pubstate = {published},
tppubtype = {article}
}
The increasing consideration of Convolutional Neural Networks (CNN) has not prevented the use of the k-Nearest Neighbor (kNN) method. In fact, a hybrid CNN-kNN approach is an interesting option in which the network specializes in feature extraction through its activations (Neural Codes), while the kNN has the advantage of performing a retrieval by means of similarity. However, this hybrid approach also has the disadvantages of the kNN search, and especially its high computational cost which is, in principle, undesirable for large-scale data. In this paper, we present the first comprehensive study of efficient kNN search algorithms using this hybrid CNN-kNN approach. This has been done by considering up to 16 different algorithms, each of which is evaluated with a different parametrization, in 7 datasets of heterogeneous composition. Our results show that no single algorithm is capable of covering all aspects, but rather that each family of algorithms is better suited to specific aspects of the problem. This signifies that Fast Similarity Search algorithms maintain their performance, but do not reduce the cost as much as the Data Reduction family does. In turn, the Approximated Similarity Search family is postulated as a good option when attempting to balance accuracy and efficiency. The experiments also suggest that considering statistical transformation algorithms such as Linear Discriminant Analysis might be useful in certain cases. Rizo, D.; Calvo-Zaragoza, J.; Iñesta, J. M.
Preservación del patrimonio musical mediante transcripción asistida por ordenador Book Chapter
In: Isusi-Fagoaga, R.; Serrano, F. Villanueva (Ed.): La música de la Corona d’Aragó: investigació, transferència i educació, Chapter 15, pp. 275–298, Universitat de València, 2020, ISBN: 978-84-09-19985-3.
@inbook{k466,
title = {Preservación del patrimonio musical mediante transcripción asistida por ordenador},
author = {D. Rizo and J. Calvo-Zaragoza and J. M. Iñesta},
editor = {R. Isusi-Fagoaga and F. Villanueva Serrano},
isbn = {978-84-09-19985-3},
year = {2020},
date = {2020-01-01},
urldate = {2020-01-01},
booktitle = {La música de la Corona d’Aragó: investigació, transferència i educació},
pages = {275--298},
publisher = {Universitat de València},
chapter = {15},
keywords = {HispaMus},
pubstate = {published},
tppubtype = {inbook}
}
2019
Calvo-Zaragoza, J.; Toselli, A. H.; Vidal, E.
Handwritten Music Recognition for Mensural Notation with Convolutional Recurrent Neural Networks Journal Article
In: Pattern Recognition Letters, vol. 128, pp. 115–121, 2019.
Links | BibTeX | Tags: HispaMus
@article{k415,
title = {Handwritten Music Recognition for Mensural Notation with Convolutional Recurrent Neural Networks},
author = {J. Calvo-Zaragoza and A. H. Toselli and E. Vidal},
url = {https://grfia.dlsi.ua.es/repositori/grfia/pubs/415/1-s2.0-S0167865519302338-main.pdf},
year = {2019},
date = {2019-12-01},
journal = {Pattern Recognition Letters},
volume = {128},
pages = {115--121},
keywords = {HispaMus},
pubstate = {published},
tppubtype = {article}
}
Iñesta, J. M.; Rizo, D.; Calvo-Zaragoza, J.
MuRET as a software for the transcription of historical archives Proceedings Article
In: Proceedings of the 2nd Workshop on Reading Music Systems, WoRMS, pp. 12–15, Delft (The Nederlands), 2019.
Abstract | Links | BibTeX | Tags: HispaMus
@inproceedings{k437,
title = {MuRET as a software for the transcription of historical archives},
author = {J. M. Iñesta and D. Rizo and J. Calvo-Zaragoza},
url = {https://grfia.dlsi.ua.es/repositori/grfia/pubs/437/worms-2019-inesta.pdf},
year = {2019},
date = {2019-11-01},
booktitle = {Proceedings of the 2nd Workshop on Reading Music Systems, WoRMS},
pages = {12--15},
address = {Delft (The Nederlands)},
abstract = {The transcription process from historical hand- written music manuscripts to a structured digital encoding has been traditionally performed following a fully manual workflow. At most, it has received some technological support in particular stages, like optical music recognition (OMR) of the source images, or transcription to modern notation with music edition applications. Currently, there is no mature and stable enough solution for the OMR problem, and the most used music editors do not support early notations, such as the mensural notation. A new tool called MUsic Recognition, Encoding, and Transcription (MuRET) has been developed, which covers all transcription phases, from the manuscript image to the encoded digital content. MuRET is designed as a machine-learning based research tool, allowing different processing approaches to be used, and producing both the expected transcribed contents in standard encodings and data for the study of the transcription process.},
keywords = {HispaMus},
pubstate = {published},
tppubtype = {inproceedings}
}
The transcription process from historical hand- written music manuscripts to a structured digital encoding has been traditionally performed following a fully manual workflow. At most, it has received some technological support in particular stages, like optical music recognition (OMR) of the source images, or transcription to modern notation with music edition applications. Currently, there is no mature and stable enough solution for the OMR problem, and the most used music editors do not support early notations, such as the mensural notation. A new tool called MUsic Recognition, Encoding, and Transcription (MuRET) has been developed, which covers all transcription phases, from the manuscript image to the encoded digital content. MuRET is designed as a machine-learning based research tool, allowing different processing approaches to be used, and producing both the expected transcribed contents in standard encodings and data for the study of the transcription process. Ríos-Vila, A.; Calvo-Zaragoza, J.; Rizo, D.; Iñesta, J. M.
ReadSco: An Open-Source Web-Based Optical Music Recognition Tool Proceedings Article
In: Calvo-Zaragoza, J.; Pacha, A. (Ed.): Proc. of the 2nd International Workshop on Reading Music Systems, WoRMS 2019, pp. 27-30, Delft (The Netherlands), 2019.
Abstract | Links | BibTeX | Tags: HispaMus
@inproceedings{k439,
title = {ReadSco: An Open-Source Web-Based Optical Music Recognition Tool},
author = {A. Ríos-Vila and J. Calvo-Zaragoza and D. Rizo and J. M. Iñesta},
editor = {J. Calvo-Zaragoza and A. Pacha},
url = {https://grfia.dlsi.ua.es/repositori/grfia/pubs/439/ReadSco_Worms2019.pdf},
year = {2019},
date = {2019-11-01},
urldate = {2019-11-01},
booktitle = {Proc. of the 2nd International Workshop on Reading Music Systems, WoRMS 2019},
pages = {27-30},
address = {Delft (The Netherlands)},
abstract = {We introduce READSCO, an open-source web-based community tool for Optical Music Recognition. READSCO aims to serve as a connection between research results and practical use. We describe the design decisions considered to both favours a rapid integration of new advances from the research field and
facilitate the community’s participation in its development. The project is still in its planning phase, so this work is a good opportunity to present the main idea and get direct feedback from other researchers.},
keywords = {HispaMus},
pubstate = {published},
tppubtype = {inproceedings}
}
We introduce READSCO, an open-source web-based community tool for Optical Music Recognition. READSCO aims to serve as a connection between research results and practical use. We describe the design decisions considered to both favours a rapid integration of new advances from the research field and
facilitate the community’s participation in its development. The project is still in its planning phase, so this work is a good opportunity to present the main idea and get direct feedback from other researchers. Iñesta, J. M.; Rizo, D.; Calvo-Zaragoza, J.
Recognition, encoding, and transcription of early mensural handwritten music Proceedings Article
In: procedings of the 13th International Workshop on Graphics Recognition (GREC 2019), pp. 13-14, Sydney (Australia), 2019.
@inproceedings{k416,
title = {Recognition, encoding, and transcription of early mensural handwritten music},
author = {J. M. Iñesta and D. Rizo and J. Calvo-Zaragoza},
year = {2019},
date = {2019-09-01},
booktitle = {procedings of the 13th International Workshop on Graphics Recognition (GREC 2019)},
pages = {13-14},
address = {Sydney (Australia)},
keywords = {HispaMus},
pubstate = {published},
tppubtype = {inproceedings}
}
Alfaro-Contreras, M.; Calvo-Zaragoza, J.; Iñesta, J. M.
Approaching End-to-end Optical Music Recognition for Homophonic Scores Proceedings Article
In: Iberian Conference on Pattern Recognition and Image Analysis, IbPRIA 2019, pp. 147–158, Springer, Madrid, 2019, ISBN: 978-3-030-31321-0.
Links | BibTeX | Tags: HispaMus
@inproceedings{k419,
title = {Approaching End-to-end Optical Music Recognition for Homophonic Scores},
author = {M. Alfaro-Contreras and J. Calvo-Zaragoza and J. M. Iñesta},
url = {https://link.springer.com/chapter/10.1007/978-3-030-31321-0_13},
isbn = {978-3-030-31321-0},
year = {2019},
date = {2019-08-01},
urldate = {2019-08-01},
booktitle = {Iberian Conference on Pattern Recognition and Image Analysis, IbPRIA 2019},
pages = {147--158},
publisher = {Springer},
address = {Madrid},
keywords = {HispaMus},
pubstate = {published},
tppubtype = {inproceedings}
}
Nuñez-Alcover, Alicia; León, P. J. Ponce; Calvo-Zaragoza, J.
Glyph and Position Classification of Music Symbols in Early Music Manuscripts Proceedings Article
In: Morales, A.; Fierrez, J.; Sánchez, J. Salvador; Ribeiro, B. (Ed.): Proc. of the 9th Iberian Conference on Pattern Recognition and Image Analysis, LNCS vol. 11867, pp. 159-168, AERFAI, APRP Springer, Madrid, Spain, 2019, ISBN: 978-3-030-31331-9.
Abstract | Links | BibTeX | Tags: HispaMus
@inproceedings{k420,
title = {Glyph and Position Classification of Music Symbols in Early Music Manuscripts},
author = {Alicia Nuñez-Alcover and P. J. Ponce León and J. Calvo-Zaragoza},
editor = {A. Morales and J. Fierrez and J. Salvador Sánchez and B. Ribeiro},
url = {https://grfia.dlsi.ua.es/repositori/grfia/pubs/420/IbPRIA2019_paper_91-1.pdf},
isbn = {978-3-030-31331-9},
year = {2019},
date = {2019-07-01},
booktitle = {Proc. of the 9th Iberian Conference on Pattern Recognition and Image Analysis, LNCS vol. 11867},
pages = {159-168},
publisher = {Springer},
address = {Madrid, Spain},
organization = {AERFAI, APRP},
abstract = {Optical Music Recognition is a field of research that automates the reading of musical scores so as to transcribe their content into a structured digital format. When dealing with music manuscripts, the traditional workflow establishes separate stages of detection and classification of musical symbols. In the latter, most of the research has focused on detecting musical glyphs, ignoring that the meaning of a musical symbol is defined by two components: its glyph and its position within the staff. In this paper we study how to perform both glyph and position classification of handwritten musical symbols in early music manuscripts written in white Mensural notation, a common notation system used for the most part of the XVI and XVII centuries. We make use of Convolutional Neural Networks as the classification method, and we tested several alternatives such as using independent models for each component, combining label spaces, or using both multi-input and multi-output models. Our results on early music manuscripts provide insights about the effectiveness and efficiency of each approach.},
keywords = {HispaMus},
pubstate = {published},
tppubtype = {inproceedings}
}
Optical Music Recognition is a field of research that automates the reading of musical scores so as to transcribe their content into a structured digital format. When dealing with music manuscripts, the traditional workflow establishes separate stages of detection and classification of musical symbols. In the latter, most of the research has focused on detecting musical glyphs, ignoring that the meaning of a musical symbol is defined by two components: its glyph and its position within the staff. In this paper we study how to perform both glyph and position classification of handwritten musical symbols in early music manuscripts written in white Mensural notation, a common notation system used for the most part of the XVI and XVII centuries. We make use of Convolutional Neural Networks as the classification method, and we tested several alternatives such as using independent models for each component, combining label spaces, or using both multi-input and multi-output models. Our results on early music manuscripts provide insights about the effectiveness and efficiency of each approach. Mateiu, T. N.; Gallego, A. J.; Calvo-Zaragoza, J.
Domain Adaptation for Handwritten Symbol Recognition: A Case of Study in Old Music Manuscripts Proceedings Article
In: Iberian Conference on Pattern Recognition and Image Analysis, IbPRIA, pp. 135–146, Springer International Publishing, Madrid, Spain, 2019, ISBN: 978-3-030-31321-0.
Abstract | BibTeX | Tags: HispaMus
@inproceedings{k410,
title = {Domain Adaptation for Handwritten Symbol Recognition: A Case of Study in Old Music Manuscripts},
author = {T. N. Mateiu and A. J. Gallego and J. Calvo-Zaragoza},
isbn = {978-3-030-31321-0},
year = {2019},
date = {2019-07-01},
urldate = {2019-07-01},
booktitle = {Iberian Conference on Pattern Recognition and Image Analysis, IbPRIA},
pages = {135--146},
publisher = {Springer International Publishing},
address = {Madrid, Spain},
abstract = {The existence of a large amount of untranscripted music manuscripts has caused initiatives that use Machine Learning (ML) for Optical Music Recognition, in order to efficiently transcribe the music sources into a machine-readable format. Although most music manuscript are similar in nature, they inevitably vary from one another. This fact can negatively influence the complexity of the classification task because most ML models fail to transfer their knowledge from one domain to another, thereby requiring learning from scratch on new domains after manually labeling new data. This work studies the ability of a Domain Adversarial Neural Network for domain adaptation in the context of classifying handwritten music symbols. The main idea is to exploit the knowledge of a specific manuscript to classify symbols from different (unlabeled) manuscripts. The reported results are promising, obtaining a substantial improvement over a conventional Convolutional Neural Network approach, which can be used as a basis for future research.},
keywords = {HispaMus},
pubstate = {published},
tppubtype = {inproceedings}
}
The existence of a large amount of untranscripted music manuscripts has caused initiatives that use Machine Learning (ML) for Optical Music Recognition, in order to efficiently transcribe the music sources into a machine-readable format. Although most music manuscript are similar in nature, they inevitably vary from one another. This fact can negatively influence the complexity of the classification task because most ML models fail to transfer their knowledge from one domain to another, thereby requiring learning from scratch on new domains after manually labeling new data. This work studies the ability of a Domain Adversarial Neural Network for domain adaptation in the context of classifying handwritten music symbols. The main idea is to exploit the knowledge of a specific manuscript to classify symbols from different (unlabeled) manuscripts. The reported results are promising, obtaining a substantial improvement over a conventional Convolutional Neural Network approach, which can be used as a basis for future research. Calvo-Zaragoza, J.; Rico-Juan, J. R.; Gallego, A. J.
Ensemble classification from deep predictions with test data augmentation Journal Article
In: Soft Computing, 2019, ISSN: 1433-7479.
@article{k409,
title = {Ensemble classification from deep predictions with test data augmentation},
author = {J. Calvo-Zaragoza and J. R. Rico-Juan and A. J. Gallego},
issn = {1433-7479},
year = {2019},
date = {2019-04-01},
urldate = {2019-04-01},
journal = {Soft Computing},
abstract = {Data augmentation has become a standard step to improve the predictive power and robustness of convolutional neural networks by means of the synthetic generation of new samples depicting different deformations. This step has been traditionally considered to improve the network at the training stage. In this work, however, we study the use of data augmentation at classification time. That is, the test sample is augmented, following the same procedure considered for training, and the decision is taken with an ensemble prediction over all these samples. We present comprehensive experimentation with several datasets and ensemble decisions, considering a rather generic data augmentation procedure. Our results show that performing this step is able to boost the original classification, even when the room for improvement is limited.},
keywords = {},
pubstate = {published},
tppubtype = {article}
}
Data augmentation has become a standard step to improve the predictive power and robustness of convolutional neural networks by means of the synthetic generation of new samples depicting different deformations. This step has been traditionally considered to improve the network at the training stage. In this work, however, we study the use of data augmentation at classification time. That is, the test sample is augmented, following the same procedure considered for training, and the decision is taken with an ensemble prediction over all these samples. We present comprehensive experimentation with several datasets and ensemble decisions, considering a rather generic data augmentation procedure. Our results show that performing this step is able to boost the original classification, even when the room for improvement is limited. Calvo-Zaragoza, J.; Gallego, A. J.
A selectional auto-encoder approach for document image binarization Journal Article
In: Pattern Recognition, vol. 86, pp. 37-47, 2019, ISSN: 0031-3203.
Abstract | Links | BibTeX | Tags: GRE16-14
@article{k395,
title = {A selectional auto-encoder approach for document image binarization},
author = {J. Calvo-Zaragoza and A. J. Gallego},
url = {https://grfia.dlsi.ua.es/repositori/grfia/pubs/395/1706.10241.pdf},
issn = {0031-3203},
year = {2019},
date = {2019-01-01},
journal = {Pattern Recognition},
volume = {86},
pages = {37-47},
abstract = {Binarization plays a key role in the automatic information retrieval from document images. This process is usually performed in the first stages of document analysis systems, and serves as a basis for subsequent steps. Hence it has to be robust in order to allow the full analysis workflow to be successful. Several methods for document image binarization have been proposed so far, most of which are based on hand-crafted image processing strategies. Recently, Convolutional Neural Networks have shown an amazing performance in many disparate duties related to computer vision. In this paper we discuss the use of convolutional auto-encoders devoted to learning an end-to-end map from an input image to its selectional output, in which activations indicate the likelihood of pixels to be either foreground or background. Once trained, documents can therefore be binarized by parsing them through the model and applying a global threshold. This approach has proven to outperform existing binarization strategies in a number of document types.},
keywords = {GRE16-14},
pubstate = {published},
tppubtype = {article}
}
Binarization plays a key role in the automatic information retrieval from document images. This process is usually performed in the first stages of document analysis systems, and serves as a basis for subsequent steps. Hence it has to be robust in order to allow the full analysis workflow to be successful. Several methods for document image binarization have been proposed so far, most of which are based on hand-crafted image processing strategies. Recently, Convolutional Neural Networks have shown an amazing performance in many disparate duties related to computer vision. In this paper we discuss the use of convolutional auto-encoders devoted to learning an end-to-end map from an input image to its selectional output, in which activations indicate the likelihood of pixels to be either foreground or background. Once trained, documents can therefore be binarized by parsing them through the model and applying a global threshold. This approach has proven to outperform existing binarization strategies in a number of document types. Rico-Juan, J. R.; Gallego, A. J.; Calvo-Zaragoza, J.
Automatic detection of inconsistencies between numerical scores and textual feedback in peer-assessment processes with machine learning Journal Article
In: Computers and Education, vol. 140, pp. 103609, 2019.
@article{k414,
title = {Automatic detection of inconsistencies between numerical scores and textual feedback in peer-assessment processes with machine learning},
author = {J. R. Rico-Juan and A. J. Gallego and J. Calvo-Zaragoza},
year = {2019},
date = {2019-01-01},
journal = {Computers and Education},
volume = {140},
pages = {103609},
abstract = {The use of peer assessment for open-ended activities has advantages for both teachers and students. Teachers might reduce the workload of the correction process and students achieve a better understanding of the subject by evaluating the activities of their peers. In order to ease the process, it is advisable to provide the students with a rubric over which performing the assessment of their peers and however, restricting themselves to provide only numerical scores is detrimental, as it prevents providing valuable feedback to others peers. Since this assessment produces two modalities of the same evaluation, namely numerical score and textual feedback, it is possible to apply automatic techniques to detect inconsistencies in the evaluation, thus minimizing the teachers' workload for supervising the whole process. This paper proposes a machine learning approach for the detection of such inconsistencies. To this end, we consider two different approaches, each of which is tested with different algorithms, in order to both evaluate the approach itself and find appropriate models to make it successful. The experiments carried out with 4 groups of students and 2 types of activities show that the proposed approach is able to yield reliable results, thus representing a valuable approach for ensuring a fair operation of the peer assessment process.},
keywords = {},
pubstate = {published},
tppubtype = {article}
}
The use of peer assessment for open-ended activities has advantages for both teachers and students. Teachers might reduce the workload of the correction process and students achieve a better understanding of the subject by evaluating the activities of their peers. In order to ease the process, it is advisable to provide the students with a rubric over which performing the assessment of their peers and however, restricting themselves to provide only numerical scores is detrimental, as it prevents providing valuable feedback to others peers. Since this assessment produces two modalities of the same evaluation, namely numerical score and textual feedback, it is possible to apply automatic techniques to detect inconsistencies in the evaluation, thus minimizing the teachers' workload for supervising the whole process. This paper proposes a machine learning approach for the detection of such inconsistencies. To this end, we consider two different approaches, each of which is tested with different algorithms, in order to both evaluate the approach itself and find appropriate models to make it successful. The experiments carried out with 4 groups of students and 2 types of activities show that the proposed approach is able to yield reliable results, thus representing a valuable approach for ensuring a fair operation of the peer assessment process. Rico-Juan, J. R.; Valero-Mas, J. J.; Calvo-Zaragoza, J.
Extensions to rank-based prototype selection in k-Nearest Neighbour classification Journal Article
In: Applied Soft Computing, no. 105803, 2019, ISSN: 1568-4946.
@article{k418,
title = {Extensions to rank-based prototype selection in k-Nearest Neighbour classification},
author = {J. R. Rico-Juan and J. J. Valero-Mas and J. Calvo-Zaragoza},
issn = {1568-4946},
year = {2019},
date = {2019-01-01},
journal = {Applied Soft Computing},
number = {105803},
abstract = {The k-nearest neighbour rule is commonly considered for classification tasks given its straightforward implementation and good performance in many applications. However, its efficiency represents an obstacle in real-case scenarios because the classification requires computing a distance to every single prototype of the training set. Prototype Selection (PS) is a typical approach to alleviate this problem, which focuses on reducing the size of the training set by selecting the most interesting prototypes. In this context, rank methods have been postulated as a good solution: following some heuristics, these methods perform an ordering of the prototypes according to their relevance in the classification task, which is then used to select the most relevant ones. This work presents a significant improvement of existing rank methods by proposing two extensions: i) a greater robustness against noise at label level by considering the parameter `k' of the classification in the selection process and and ii) a new parameter-free rule to select the prototypes once they have been ordered. The experiments performed in different scenarios and datasets demonstrate the goodness of these extensions. Also, it is reported that the new full approach is competitive with respect to existing PS algorithms.},
keywords = {},
pubstate = {published},
tppubtype = {article}
}
The k-nearest neighbour rule is commonly considered for classification tasks given its straightforward implementation and good performance in many applications. However, its efficiency represents an obstacle in real-case scenarios because the classification requires computing a distance to every single prototype of the training set. Prototype Selection (PS) is a typical approach to alleviate this problem, which focuses on reducing the size of the training set by selecting the most interesting prototypes. In this context, rank methods have been postulated as a good solution: following some heuristics, these methods perform an ordering of the prototypes according to their relevance in the classification task, which is then used to select the most relevant ones. This work presents a significant improvement of existing rank methods by proposing two extensions: i) a greater robustness against noise at label level by considering the parameter `k' of the classification in the selection process and and ii) a new parameter-free rule to select the prototypes once they have been ordered. The experiments performed in different scenarios and datasets demonstrate the goodness of these extensions. Also, it is reported that the new full approach is competitive with respect to existing PS algorithms. Baró, A.; Riba, P.; Calvo-Zaragoza, J.; Fornés, A.
From Optical Music Recognition to Handwritten Music Recognition: A baseline Journal Article
In: Pattern Recognition Letters, vol. 123, pp. 1-8, 2019.
BibTeX | Tags:
@article{k428,
title = {From Optical Music Recognition to Handwritten Music Recognition: A baseline},
author = {A. Baró and P. Riba and J. Calvo-Zaragoza and A. Fornés},
year = {2019},
date = {2019-01-01},
journal = {Pattern Recognition Letters},
volume = {123},
pages = {1-8},
keywords = {},
pubstate = {published},
tppubtype = {article}
}
Román, M. A.; Pertusa, A.; Calvo-Zaragoza, J.
A Holistic Approach to Polyphonic Music Transcription with Neural Networks Proceedings Article
In: Flexer, Julián Urbano Geoffroy Peeters Arthur (Ed.): Proc. of the 20th International Society for Music Information Retrieval Conference, ISMIR, pp. 731–737, Delf, The Netherlands, 2019.
@inproceedings{k442,
title = {A Holistic Approach to Polyphonic Music Transcription with Neural Networks},
author = {M. A. Román and A. Pertusa and J. Calvo-Zaragoza},
editor = {Julián Urbano Geoffroy Peeters Arthur Flexer},
year = {2019},
date = {2019-01-01},
booktitle = {Proc. of the 20th International Society for Music Information Retrieval Conference, ISMIR},
pages = {731--737},
address = {Delf, The Netherlands},
keywords = {HispaMus},
pubstate = {published},
tppubtype = {inproceedings}
}
Garay-Maestre, U.; Gallego, A. J.; Calvo-Zaragoza, J.
Data Augmentation via Variational Auto-Encoders Proceedings Article
In: Progress in Pattern Recognition, Image Analysis, Computer Vision, and Applications, pp. 29–37, 2019, ISBN: 978-3-030-13469-3.
Abstract | BibTeX | Tags: HispaMus
@inproceedings{k406,
title = {Data Augmentation via Variational Auto-Encoders},
author = {U. Garay-Maestre and A. J. Gallego and J. Calvo-Zaragoza},
isbn = {978-3-030-13469-3},
year = {2019},
date = {2019-01-01},
urldate = {2019-01-01},
booktitle = {Progress in Pattern Recognition, Image Analysis, Computer Vision, and Applications},
pages = {29--37},
abstract = {Data augmentation is a widely considered technique to improve the performance of Convolutional Neural Networks during training. This step consists in synthetically generate new labeled data by perturbing the samples of the training set, which is expected to provide more robustness to the learning process. The problem is that the augmentation procedure has to be adjusted manually because the perturbations considered must make sense for the task at issue. In this paper we propose the use of Variational Auto-Encoders (VAEs) to generate new synthetic samples, instead of resorting to heuristic strategies. VAEs are powerful generative models that learn a parametric latent space of the input domain from which new samples can be generated. In our experiments over the well-known MNIST dataset, the data augmentation by VAEs improves the base results, yet to a lesser extent of that obtained by a well-adjusted conventional data augmentation. However, the combination of both conventional and VAE-guided data augmentations outperforms all the results, thereby demonstrating the goodness of our proposal.},
keywords = {HispaMus},
pubstate = {published},
tppubtype = {inproceedings}
}
Data augmentation is a widely considered technique to improve the performance of Convolutional Neural Networks during training. This step consists in synthetically generate new labeled data by perturbing the samples of the training set, which is expected to provide more robustness to the learning process. The problem is that the augmentation procedure has to be adjusted manually because the perturbations considered must make sense for the task at issue. In this paper we propose the use of Variational Auto-Encoders (VAEs) to generate new synthetic samples, instead of resorting to heuristic strategies. VAEs are powerful generative models that learn a parametric latent space of the input domain from which new samples can be generated. In our experiments over the well-known MNIST dataset, the data augmentation by VAEs improves the base results, yet to a lesser extent of that obtained by a well-adjusted conventional data augmentation. However, the combination of both conventional and VAE-guided data augmentations outperforms all the results, thereby demonstrating the goodness of our proposal.2018
Rico-Juan, J. R.; Gallego, A. J.; Valero-Mas, J. J.; Calvo-Zaragoza, J.
Statistical semi-supervised system for grading multiple peer-reviewed open-ended works Journal Article
In: Computers & Education, vol. 126, no. 1, pp. 264-282, 2018.
Abstract | Links | BibTeX | Tags:
@article{k394,
title = {Statistical semi-supervised system for grading multiple peer-reviewed open-ended works},
author = {J. R. Rico-Juan and A. J. Gallego and J. J. Valero-Mas and J. Calvo-Zaragoza},
url = {https://grfia.dlsi.ua.es/repositori/grfia/pubs/394/statistical-semi-supervised.pdf},
year = {2018},
date = {2018-11-01},
journal = {Computers & Education},
volume = {126},
number = {1},
pages = {264-282},
abstract = {In the education context, open-ended works generally entail a series of benefits as the possibility of develop original ideas and a more productive learning process to the student rather than closed-answer activities. Nevertheless, such works suppose a significant correction workload to the teacher in contrast to the latter ones that can be self-corrected. Furthermore, such workload turns to be intractable with large groups of students. In order to maintain the advantages of open-ended works with a reasonable amount of correction effort, this article proposes a novel methodology: students perform the corrections using a rubric (closed Likert scale) as a guideline in a peer-review fashion and then, their markings are automatically analyzed with statistical tools to detect possible biased scorings and finally, in the event the statistical analysis detects a biased case, the teacher is required to intervene to manually correct the assignment. This methodology has been tested on two different assignments with two heterogeneous groups of people to assess the robustness and reliability of the proposal. As a result, we obtain values over 95% in the confidence of the intra-class correlation test (ICC) between the grades computed by our proposal and those directly resulting from the manual correction of the teacher. These figures confirm that the evaluation obtained with the proposed methodology is statistically similar to that of the manual correction of the teacher with a remarkable decrease in terms of effort.},
keywords = {},
pubstate = {published},
tppubtype = {article}
}
In the education context, open-ended works generally entail a series of benefits as the possibility of develop original ideas and a more productive learning process to the student rather than closed-answer activities. Nevertheless, such works suppose a significant correction workload to the teacher in contrast to the latter ones that can be self-corrected. Furthermore, such workload turns to be intractable with large groups of students. In order to maintain the advantages of open-ended works with a reasonable amount of correction effort, this article proposes a novel methodology: students perform the corrections using a rubric (closed Likert scale) as a guideline in a peer-review fashion and then, their markings are automatically analyzed with statistical tools to detect possible biased scorings and finally, in the event the statistical analysis detects a biased case, the teacher is required to intervene to manually correct the assignment. This methodology has been tested on two different assignments with two heterogeneous groups of people to assess the robustness and reliability of the proposal. As a result, we obtain values over 95% in the confidence of the intra-class correlation test (ICC) between the grades computed by our proposal and those directly resulting from the manual correction of the teacher. These figures confirm that the evaluation obtained with the proposed methodology is statistically similar to that of the manual correction of the teacher with a remarkable decrease in terms of effort. Rizo, D.; Calvo-Zaragoza, J.; Iñesta, J. M.
MuRET: a music recognition, encoding, and transcription tool Proceedings Article
In: Proceedings of the 5th International Conference on Digital Libraries for Musicology (DLfM'18), pp. 52–56, ACM, París, France, 2018, ISBN: 978-1-4503-6522-2.
Abstract | BibTeX | Tags: HispaMus
@inproceedings{k397,
title = {MuRET: a music recognition, encoding, and transcription tool},
author = {D. Rizo and J. Calvo-Zaragoza and J. M. Iñesta},
isbn = {978-1-4503-6522-2},
year = {2018},
date = {2018-09-01},
booktitle = {Proceedings of the 5th International Conference on Digital Libraries for Musicology (DLfM'18)},
pages = {52--56},
publisher = {ACM},
address = {París, France},
abstract = {The transcription process from early and modern notation ma-nu-scripts to a structured digital encoding has been traditionally performed following a fully manual workflow. At most it has received some technological support in particular stages, like optical music recognition (OMR) of the source images, or transcription to modern notation with music edition applications. Currently, there is no mature and stable enough solution for the OMR problem, and the most used music editors do not support early notations, such as the mensural one. In this work, a new tool called MUsic Recognition, Encoding, and Transcription (MuRET) is introduced, which covers all transcription phases, from the manuscript source to the encoded digital content. MuRET is designed as a technology-focused research tool, allowing different processing approaches to be used, and producing both the expected transcribed contents in standard encodings and data for the study of the transcription process itself.},
keywords = {HispaMus},
pubstate = {published},
tppubtype = {inproceedings}
}
The transcription process from early and modern notation ma-nu-scripts to a structured digital encoding has been traditionally performed following a fully manual workflow. At most it has received some technological support in particular stages, like optical music recognition (OMR) of the source images, or transcription to modern notation with music edition applications. Currently, there is no mature and stable enough solution for the OMR problem, and the most used music editors do not support early notations, such as the mensural one. In this work, a new tool called MUsic Recognition, Encoding, and Transcription (MuRET) is introduced, which covers all transcription phases, from the manuscript source to the encoded digital content. MuRET is designed as a technology-focused research tool, allowing different processing approaches to be used, and producing both the expected transcribed contents in standard encodings and data for the study of the transcription process itself. Castellanos, F. J.; Calvo-Zaragoza, J.; Vigliensoni, G.; Fujinaga, I.
Document Analysis of Music Score Images with Selectional Auto-Encoders Proceedings Article
In: Proceedings of the19th International Society for Music Information Retrieval Conference, pp. 256-263, ISMIR 2018 Paris, France, 2018.
@inproceedings{k449,
title = {Document Analysis of Music Score Images with Selectional Auto-Encoders},
author = {F. J. Castellanos and J. Calvo-Zaragoza and G. Vigliensoni and I. Fujinaga},
year = {2018},
date = {2018-09-01},
booktitle = {Proceedings of the19th International Society for Music Information Retrieval Conference},
pages = {256-263},
address = {Paris, France},
organization = {ISMIR 2018},
keywords = {HispaMus},
pubstate = {published},
tppubtype = {inproceedings}
}
Calvo-Zaragoza, J.; Rizo, D.
Camera-PrIMuS: Neural End-to-End Optical Music Recognition on Realistic Monophonic Scores Proceedings Article
In: Proceedings of the 19th International Society of Music Information Retrieval (ISMIR), International Society of Music Information Retrieval 2018.
Links | BibTeX | Tags: HispaMus
@inproceedings{k391,
title = {Camera-PrIMuS: Neural End-to-End Optical Music Recognition on Realistic Monophonic Scores},
author = {J. Calvo-Zaragoza and D. Rizo},
url = {https://grfia.dlsi.ua.es/repositori/grfia/pubs/391/33_Paper.pdf},
year = {2018},
date = {2018-09-01},
urldate = {2018-09-01},
booktitle = {Proceedings of the 19th International Society of Music Information Retrieval (ISMIR)},
organization = {International Society of Music Information Retrieval},
keywords = {HispaMus},
pubstate = {published},
tppubtype = {inproceedings}
}
Román, M. A.; Pertusa, A.; Calvo-Zaragoza, J.
An End-to-End Framework for Audio-to-Score Music Transcription on Monophonic Excerpts Proceedings Article
In: Proc. of the 19th International Society for Music Information Retrieval Conference (ISMIR), Paris, France, 2018.
BibTeX | Tags: GRE16-14, HispaMus
@inproceedings{k389,
title = {An End-to-End Framework for Audio-to-Score Music Transcription on Monophonic Excerpts},
author = {M. A. Román and A. Pertusa and J. Calvo-Zaragoza},
year = {2018},
date = {2018-09-01},
urldate = {2018-09-01},
booktitle = {Proc. of the 19th International Society for Music Information Retrieval Conference (ISMIR)},
address = {Paris, France},
keywords = {GRE16-14, HispaMus},
pubstate = {published},
tppubtype = {inproceedings}
}
Iñesta, J. M.; León, P. J. Ponce; Rizo, D.; Oncina, J.; Micó, L.; Rico-Juan, J. R.; Pérez-Sancho, C.; Pertusa, A.
HISPAMUS: Handwritten Spanish Music Heritage Preservation by Automatic Transcription Proceedings Article
In: Calvo-Zaragoza, J.; Jr., J. Hajic; Pacha, A. (Ed.): Proceedings of the 1st Workshop on Reading Music Systems, WoRMS, pp. 17–18, Paris, 2018.
Abstract | Links | BibTeX | Tags: HispaMus
@inproceedings{k396,
title = {HISPAMUS: Handwritten Spanish Music Heritage Preservation by Automatic Transcription},
author = {J. M. Iñesta and P. J. Ponce León and D. Rizo and J. Oncina and L. Micó and J. R. Rico-Juan and C. Pérez-Sancho and A. Pertusa},
editor = {J. Calvo-Zaragoza and J. Hajic Jr. and A. Pacha},
url = {https://grfia.dlsi.ua.es/repositori/grfia/pubs/396/worms-2018-inesta.pdf},
year = {2018},
date = {2018-09-01},
urldate = {2018-09-01},
booktitle = {Proceedings of the 1st Workshop on Reading Music Systems, WoRMS},
pages = {17--18},
address = {Paris},
abstract = {The HISPAMUS proposal aims at enhancing the Hispanic music heritage from the 15th to the 19th centuries, by exploiting the digital resources of these collections. In addition, thousands of oral tradition melodies that were compiled by folklorists in the 1950s decade are digitized just as images, currently without the possibility of content-based search or study. It is necessary to develop services and tools for the benefit of archives, libraries, scholars, computer scientists and general public. HISPAMUS tries to provide smart access to archival manuscripts of music scores, allowing its reuse and exploitation. In order to reach this ambitious goal, our group can provide cutting-edge technology in the fields of Machine Learning, Pattern Recognition, and Optical Music Recognition.},
keywords = {HispaMus},
pubstate = {published},
tppubtype = {inproceedings}
}
The HISPAMUS proposal aims at enhancing the Hispanic music heritage from the 15th to the 19th centuries, by exploiting the digital resources of these collections. In addition, thousands of oral tradition melodies that were compiled by folklorists in the 1950s decade are digitized just as images, currently without the possibility of content-based search or study. It is necessary to develop services and tools for the benefit of archives, libraries, scholars, computer scientists and general public. HISPAMUS tries to provide smart access to archival manuscripts of music scores, allowing its reuse and exploitation. In order to reach this ambitious goal, our group can provide cutting-edge technology in the fields of Machine Learning, Pattern Recognition, and Optical Music Recognition. Calvo-Zaragoza, J.; Rizo, D.
End-to-End Neural Optical Music Recognition of Monophonic Scores Journal Article
In: Applied Sciences, vol. 8, no. 4, pp. 606–623, 2018, ISSN: 2076-3417.
@article{k390,
title = {End-to-End Neural Optical Music Recognition of Monophonic Scores},
author = {J. Calvo-Zaragoza and D. Rizo},
issn = {2076-3417},
year = {2018},
date = {2018-04-01},
journal = {Applied Sciences},
volume = {8},
number = {4},
pages = {606--623},
keywords = {HispaMus},
pubstate = {published},
tppubtype = {article}
}
Castellanos, F. J.; Valero-Mas, J. J.; Calvo-Zaragoza, J.; Rico-Juan, J. R.
Oversampling imbalanced data in the string space Journal Article
In: Pattern Recognition Letters, vol. 103, pp. 32–38, 2018, ISSN: 0167-8655.
Abstract | BibTeX | Tags: GRE16-14
@article{k382,
title = {Oversampling imbalanced data in the string space},
author = {F. J. Castellanos and J. J. Valero-Mas and J. Calvo-Zaragoza and J. R. Rico-Juan},
issn = {0167-8655},
year = {2018},
date = {2018-02-01},
journal = {Pattern Recognition Letters},
volume = {103},
pages = {32--38},
abstract = {Imbalanced data is a typical problem in the supervised classification field, which occurs when the different classes are not equally represented. This fact typically results in the classifier biasing its performance towards the class representing the majority of the elements. Many methods have been proposed to alleviate this scenario, yet all of them assume that data is represented as feature vectors. In this paper we propose a strategy to balance a dataset whose samples are encoded as strings. Our approach is based on adapting the well-known Synthetic Minority Over-sampling Technique (SMOTE) algorithm to the string space. More precisely, data generation is achieved with an iterative approach to create artificial strings within the segment between two given samples of the training set. Results with several datasets and imbalance ratios show that the proposed strategy properly deals with the problem in all cases considered.},
keywords = {GRE16-14},
pubstate = {published},
tppubtype = {article}
}
Imbalanced data is a typical problem in the supervised classification field, which occurs when the different classes are not equally represented. This fact typically results in the classifier biasing its performance towards the class representing the majority of the elements. Many methods have been proposed to alleviate this scenario, yet all of them assume that data is represented as feature vectors. In this paper we propose a strategy to balance a dataset whose samples are encoded as strings. Our approach is based on adapting the well-known Synthetic Minority Over-sampling Technique (SMOTE) algorithm to the string space. More precisely, data generation is achieved with an iterative approach to create artificial strings within the segment between two given samples of the training set. Results with several datasets and imbalance ratios show that the proposed strategy properly deals with the problem in all cases considered. Gallego, A. J.; Pertusa, A.; Calvo-Zaragoza, J.
Improving Convolutional Neural Networks’ Accuracy in Noisy Environments Using k-Nearest Neighbors Journal Article
In: Applied Sciences, vol. 8, no. 11, 2018, ISSN: 2076-3417.
Abstract | BibTeX | Tags: HispaMus
@article{k399,
title = {Improving Convolutional Neural Networks’ Accuracy in Noisy Environments Using k-Nearest Neighbors},
author = {A. J. Gallego and A. Pertusa and J. Calvo-Zaragoza},
issn = {2076-3417},
year = {2018},
date = {2018-01-01},
journal = {Applied Sciences},
volume = {8},
number = {11},
abstract = {We present a hybrid approach to improve the accuracy of Convolutional Neural Networks (CNN) without retraining the model. The proposed architecture replaces the softmax layer by a k-Nearest Neighbor (kNN) algorithm for inference. Although this is a common technique in transfer learning, we apply it to the same domain for which the network was trained. Previous works show that neural codes (neuron activations of the last hidden layers) can benefit from the inclusion of classifiers such as support vector machines or random forests. In this work, our proposed hybrid CNN + kNN architecture is evaluated using several image datasets, network topologies and label noise levels. The results show significant accuracy improvements in the inference stage with respect to the standard CNN with noisy labels, especially with relatively large datasets such as CIFAR100. We also verify that applying the ℓ and 2
norm on neural codes is statistically beneficial for this approach.},
keywords = {HispaMus},
pubstate = {published},
tppubtype = {article}
}
We present a hybrid approach to improve the accuracy of Convolutional Neural Networks (CNN) without retraining the model. The proposed architecture replaces the softmax layer by a k-Nearest Neighbor (kNN) algorithm for inference. Although this is a common technique in transfer learning, we apply it to the same domain for which the network was trained. Previous works show that neural codes (neuron activations of the last hidden layers) can benefit from the inclusion of classifiers such as support vector machines or random forests. In this work, our proposed hybrid CNN + kNN architecture is evaluated using several image datasets, network topologies and label noise levels. The results show significant accuracy improvements in the inference stage with respect to the standard CNN with noisy labels, especially with relatively large datasets such as CIFAR100. We also verify that applying the ℓ and 2
norm on neural codes is statistically beneficial for this approach. Sober-Mira, J.; Calvo-Zaragoza, J.; Rizo, D.; Iñesta, J. M.
Pen-Based Music Document Transcription with Convolutional Neural Networks Book Chapter
In: Fornés, A.; Lamiroy, B. (Ed.): Graphics Recognition. Current Trends and Evolutions, Chapter 6, pp. 71–80, Springer, 2018, ISBN: 978-3-030-02284-6.
Abstract | BibTeX | Tags: HispaMus
@inbook{k400,
title = {Pen-Based Music Document Transcription with Convolutional Neural Networks},
author = {J. Sober-Mira and J. Calvo-Zaragoza and D. Rizo and J. M. Iñesta},
editor = {A. Fornés and B. Lamiroy},
isbn = {978-3-030-02284-6},
year = {2018},
date = {2018-01-01},
booktitle = {Graphics Recognition. Current Trends and Evolutions},
pages = {71--80},
publisher = {Springer},
chapter = {6},
abstract = {The transcription of music sources requires new ways of interacting with musical documents. Assuming that au- tomatic technologies will never guarantee a perfect transcription, our intention is to develop an interactive system in which user and software collaborate to complete the task. Since the use of traditional software for score edition might be tedious, our work studies the interaction by means of electronic pen (e-pen). In our framework, users trace symbols using an e-pen over a digital surface, which provides both the underlying image (offline data) and the drawing made (online data). Using both sources, the system is capable of reaching an error below 4% when recognizing the symbols with a Convolutional Neural Network.},
keywords = {HispaMus},
pubstate = {published},
tppubtype = {inbook}
}
The transcription of music sources requires new ways of interacting with musical documents. Assuming that au- tomatic technologies will never guarantee a perfect transcription, our intention is to develop an interactive system in which user and software collaborate to complete the task. Since the use of traditional software for score edition might be tedious, our work studies the interaction by means of electronic pen (e-pen). In our framework, users trace symbols using an e-pen over a digital surface, which provides both the underlying image (offline data) and the drawing made (online data). Using both sources, the system is capable of reaching an error below 4% when recognizing the symbols with a Convolutional Neural Network. Calvo-Zaragoza, J.; Castellanos, F. J.; Vigliensoni, G.; Fujinaga, I.
Deep Neural Networks for Document Processing of Music Score Images Journal Article
In: Applied Sciences, vol. 8, no. 5, pp. 654, 2018.
@article{k427,
title = {Deep Neural Networks for Document Processing of Music Score Images},
author = {J. Calvo-Zaragoza and F. J. Castellanos and G. Vigliensoni and I. Fujinaga},
year = {2018},
date = {2018-01-01},
urldate = {2018-01-01},
journal = {Applied Sciences},
volume = {8},
number = {5},
pages = {654},
keywords = {HispaMus},
pubstate = {published},
tppubtype = {article}
}
Gallego, A. J.; Calvo-Zaragoza, J.; Valero-Mas, J. J.; Rico-Juan, J. R.
Clustering-based k-nearest neighbor classification for large-scale data with neural codes representation Journal Article
In: Pattern Recognition, vol. 74, pp. 531-543, 2018.
@article{k378,
title = {Clustering-based k-nearest neighbor classification for large-scale data with neural codes representation},
author = {A. J. Gallego and J. Calvo-Zaragoza and J. J. Valero-Mas and J. R. Rico-Juan},
year = {2018},
date = {2018-01-01},
urldate = {2018-01-01},
journal = {Pattern Recognition},
volume = {74},
pages = {531-543},
keywords = {TIMuL},
pubstate = {published},
tppubtype = {article}
}
2017
Sober-Mira, J.; Calvo-Zaragoza, J.; Rizo, D.; Iñesta, J. M.
Pen-based music document transcription Proceedings Article
In: Proceedings of GREC 2017, pp. 21–22, IEEE computer society, Kyoto (Japan), 2017, ISBN: 978-1-5386-3586-5.
@inproceedings{k381,
title = {Pen-based music document transcription},
author = {J. Sober-Mira and J. Calvo-Zaragoza and D. Rizo and J. M. Iñesta},
url = {https://grfia.dlsi.ua.es/repositori/grfia/pubs/381/3586c021.pdf},
isbn = {978-1-5386-3586-5},
year = {2017},
date = {2017-11-01},
booktitle = {Proceedings of GREC 2017},
pages = {21--22},
publisher = {IEEE computer society},
address = {Kyoto (Japan)},
keywords = {TIMuL},
pubstate = {published},
tppubtype = {inproceedings}
}
Calvo-Zaragoza, J.; Gallego, A. J.; Pertusa, A.
Recognition of Handwritten Music Symbols with Convolutional Neural Codes Proceedings Article
In: 14th IAPR International Conference on Document Analysis and Recognition (ICDAR), pp. 691–696, Kyoto, Japan, 2017.
BibTeX | Tags: GRE16-14, TIMuL
@inproceedings{k376,
title = {Recognition of Handwritten Music Symbols with Convolutional Neural Codes},
author = {J. Calvo-Zaragoza and A. J. Gallego and A. Pertusa},
year = {2017},
date = {2017-11-01},
urldate = {2017-11-01},
booktitle = {14th IAPR International Conference on Document Analysis and Recognition (ICDAR)},
pages = {691--696},
address = {Kyoto, Japan},
keywords = {GRE16-14, TIMuL},
pubstate = {published},
tppubtype = {inproceedings}
}
Calvo-Zaragoza, J.; Valero-Mas, J. J.; Pertusa, A
End-To-End Optical Music Recognition using Neural Networks Proceedings Article
In: Proc. of International Society for Music Information Retrieval Conference (ISMIR), Suzhou, China, 2017.
Abstract | BibTeX | Tags: GRE16-14, TIMuL
@inproceedings{k374,
title = {End-To-End Optical Music Recognition using Neural Networks},
author = {J. Calvo-Zaragoza and J. J. Valero-Mas and A Pertusa},
year = {2017},
date = {2017-10-01},
booktitle = {Proc. of International Society for Music Information Retrieval Conference (ISMIR)},
address = {Suzhou, China},
abstract = {This work addresses the Optical Music Recognition (OMR) task in an end-to-end fashion using neural net- works. The proposed architecture is based on a Recurrent Convolutional Neural Network topology that takes as input an image of a monophonic score and retrieves a sequence of music symbols as output. In the first stage, a series of convolutional filters are trained to extract meaningful fea- tures of the input image, and then a recurrent block models the sequential nature of music. The system is trained us- ing a Connectionist Temporal Classification loss function, which avoids the need for a frame-by-frame alignment be- tween the image and the ground-truth music symbols. Ex- perimentation has been carried on a set of 90,000 synthetic monophonic music scores with more than 50 different pos- sible labels. Results obtained depict classification error rates around 2 % at symbol level, thus proving the po- tential of the proposed end-to-end architecture for OMR. The source code, dataset, and trained models are publicly released for reproducible research and future comparison purposes.},
keywords = {GRE16-14, TIMuL},
pubstate = {published},
tppubtype = {inproceedings}
}
This work addresses the Optical Music Recognition (OMR) task in an end-to-end fashion using neural net- works. The proposed architecture is based on a Recurrent Convolutional Neural Network topology that takes as input an image of a monophonic score and retrieves a sequence of music symbols as output. In the first stage, a series of convolutional filters are trained to extract meaningful fea- tures of the input image, and then a recurrent block models the sequential nature of music. The system is trained us- ing a Connectionist Temporal Classification loss function, which avoids the need for a frame-by-frame alignment be- tween the image and the ground-truth music symbols. Ex- perimentation has been carried on a set of 90,000 synthetic monophonic music scores with more than 50 different pos- sible labels. Results obtained depict classification error rates around 2 % at symbol level, thus proving the po- tential of the proposed end-to-end architecture for OMR. The source code, dataset, and trained models are publicly released for reproducible research and future comparison purposes. Sober-Mira, J.; Calvo-Zaragoza, J.; Rizo, D.; Iñesta, J. M.
Multimodal Recognition for Music Document Transcription Proceedings Article
In: Proceedings of MML 2017, pp. 67–72, Barcelona, 2017.
@inproceedings{k380,
title = {Multimodal Recognition for Music Document Transcription},
author = {J. Sober-Mira and J. Calvo-Zaragoza and D. Rizo and J. M. Iñesta},
url = {https://grfia.dlsi.ua.es/repositori/grfia/pubs/380/mml17proceedings-67.pdf},
year = {2017},
date = {2017-10-01},
booktitle = {Proceedings of MML 2017},
pages = {67--72},
address = {Barcelona},
keywords = {},
pubstate = {published},
tppubtype = {inproceedings}
}
Valero-Mas, J. J.; Calvo-Zaragoza, J.; Rico-Juan, J. R.; Iñesta, J. M.
A study of Prototype Selection algorithms for Nearest Neighbour in class-imbalanced problems Proceedings Article
In: Alexandre, J. S. Sánchez L. A.; Rodrigues, J. M. F. (Ed.): Proceedings of the 8th Iberian Conference on Pattern Recognition and Image Analysis (IbPRIA), pp. 335–343, Springer, Faro, Portugal, 2017, ISBN: 978-3-319-58837-7.
@inproceedings{k362,
title = {A study of Prototype Selection algorithms for Nearest Neighbour in class-imbalanced problems},
author = {J. J. Valero-Mas and J. Calvo-Zaragoza and J. R. Rico-Juan and J. M. Iñesta},
editor = {J. S. Sánchez L. A. Alexandre and J. M. F. Rodrigues},
url = {https://grfia.dlsi.ua.es/repositori/grfia/pubs/362/CameraReady.pdf},
isbn = {978-3-319-58837-7},
year = {2017},
date = {2017-06-01},
booktitle = {Proceedings of the 8th Iberian Conference on Pattern Recognition and Image Analysis (IbPRIA)},
pages = {335--343},
publisher = {Springer},
address = {Faro, Portugal},
keywords = {TIMuL},
pubstate = {published},
tppubtype = {inproceedings}
}
Rizo, D.; Calvo-Zaragoza, J.; Iñesta, J. M.; Fujinaga, I.
About agnostic representation of musical documents for Optical Music Recognition Proceedings Article
In: Music Encoding Conference, Tours, 2017, 2017.
@inproceedings{k369,
title = {About agnostic representation of musical documents for Optical Music Recognition},
author = {D. Rizo and J. Calvo-Zaragoza and J. M. Iñesta and I. Fujinaga},
year = {2017},
date = {2017-05-01},
booktitle = {Music Encoding Conference, Tours, 2017},
keywords = {TIMuL},
pubstate = {published},
tppubtype = {inproceedings}
}
Calvo-Zaragoza, J.; Oncina, J.
Recognition of Pen-based Music Notation with Finite-State Machines Journal Article
In: Expert Systems With Applications, vol. 72, pp. 395-406, 2017.
@article{k358,
title = {Recognition of Pen-based Music Notation with Finite-State Machines},
author = {J. Calvo-Zaragoza and J. Oncina},
url = {https://grfia.dlsi.ua.es/repositori/grfia/pubs/358/recognition-pen-based.pdf},
year = {2017},
date = {2017-04-01},
journal = {Expert Systems With Applications},
volume = {72},
pages = {395-406},
keywords = {TIMuL},
pubstate = {published},
tppubtype = {article}
}
Calvo-Zaragoza, J.; Oncina, J.
An efficient approach for Interactive Sequential Pattern Recognition Journal Article
In: Pattern Recognition, vol. 64, pp. 295-304, 2017.
@article{k359,
title = {An efficient approach for Interactive Sequential Pattern Recognition},
author = {J. Calvo-Zaragoza and J. Oncina},
url = {https://grfia.dlsi.ua.es/repositori/grfia/pubs/359/efficient-approach-sequential.pdf},
year = {2017},
date = {2017-04-01},
journal = {Pattern Recognition},
volume = {64},
pages = {295-304},
keywords = {TIMuL},
pubstate = {published},
tppubtype = {article}
}
Valero-Mas, J. J.; Calvo-Zaragoza, J.; Rico-Juan, J. R.; Iñesta, J. M.
An Experimental Study on Rank Methods for Prototype Selection Journal Article
In: Soft Computing, vol. 21, no. 19, pp. 5703-–5715, 2017, ISSN: 1432-7643.
Abstract | Links | BibTeX | Tags: TIMuL
@article{k339,
title = {An Experimental Study on Rank Methods for Prototype Selection},
author = {J. J. Valero-Mas and J. Calvo-Zaragoza and J. R. Rico-Juan and J. M. Iñesta},
url = {https://grfia.dlsi.ua.es/repositori/grfia/pubs/339/SOCO-RankMethods-2016.pdf},
issn = {1432-7643},
year = {2017},
date = {2017-01-01},
journal = {Soft Computing},
volume = {21},
number = {19},
pages = {5703-–5715},
abstract = {Prototype selection is one of the most popular approaches for addressing the low efficiency issue typically found in the well-known k-Nearest Neighbour classification rule. These techniques select a representative subset from an original collection of prototypes with the premise of main- taining the same classification accuracy. Most recently, rank methods have been proposed as an alternative to develop new selection strategies. Following a certain heuristic, these methods sort the elements of the initial collection accord- ing to their relevance and then select the best possible subset by means of a parameter representing the amount of data to maintain. Due to the relative novelty of these methods, their performance and competitiveness against other strategies is still unclear. This work performs an exhaustive experimental study of such methods for prototype selection. A represen- tative collection of both classic and sophisticated algorithms are compared to the aforementioned techniques in a num- ber of datasets, including different levels of induced noise. Results report the remarkable competitiveness of these rank methods as well as their excellent trade-off between proto- type reduction and achieved accuracy.},
keywords = {TIMuL},
pubstate = {published},
tppubtype = {article}
}
Prototype selection is one of the most popular approaches for addressing the low efficiency issue typically found in the well-known k-Nearest Neighbour classification rule. These techniques select a representative subset from an original collection of prototypes with the premise of main- taining the same classification accuracy. Most recently, rank methods have been proposed as an alternative to develop new selection strategies. Following a certain heuristic, these methods sort the elements of the initial collection accord- ing to their relevance and then select the best possible subset by means of a parameter representing the amount of data to maintain. Due to the relative novelty of these methods, their performance and competitiveness against other strategies is still unclear. This work performs an exhaustive experimental study of such methods for prototype selection. A represen- tative collection of both classic and sophisticated algorithms are compared to the aforementioned techniques in a num- ber of datasets, including different levels of induced noise. Results report the remarkable competitiveness of these rank methods as well as their excellent trade-off between proto- type reduction and achieved accuracy. Calvo-Zaragoza, J.; Vigliensoni, G.; Fujinaga, I.
A machine learning framework for the categorization of elements in images of musical documents Proceedings Article
In: Proceedings of the Third International Conference on Technologies for Music Notation and Representation, 2017.
@inproceedings{k360,
title = {A machine learning framework for the categorization of elements in images of musical documents},
author = {J. Calvo-Zaragoza and G. Vigliensoni and I. Fujinaga},
url = {https://grfia.dlsi.ua.es/repositori/grfia/pubs/360/tenor-unified-categorization.pdf},
year = {2017},
date = {2017-01-01},
booktitle = {Proceedings of the Third International Conference on Technologies for Music Notation and Representation},
keywords = {TIMuL},
pubstate = {published},
tppubtype = {inproceedings}
}
Calvo-Zaragoza, J.; Vigliensoni, G.; Fujinaga, I.
Pixel-wise Binarization of Musical Documents with Convolutional Neural Networks Proceedings Article
In: Proceedings of the 15th IAPR International Conference on Machine Vision Applications, 2017.
@inproceedings{k361,
title = {Pixel-wise Binarization of Musical Documents with Convolutional Neural Networks},
author = {J. Calvo-Zaragoza and G. Vigliensoni and I. Fujinaga},
url = {https://grfia.dlsi.ua.es/repositori/grfia/pubs/361/pixel-wise-binarization.pdf},
year = {2017},
date = {2017-01-01},
booktitle = {Proceedings of the 15th IAPR International Conference on Machine Vision Applications},
keywords = {TIMuL},
pubstate = {published},
tppubtype = {inproceedings}
}
Calvo-Zaragoza, J.; Pertusa, A.; Oncina, J.
Staff-line detection and removal using a convolutional neural network Journal Article
In: Machine Vision and Applications, pp. 1-10, 2017, ISSN: 1432-1769.
Abstract | BibTeX | Tags: TIMuL
@article{k365,
title = {Staff-line detection and removal using a convolutional neural network},
author = {J. Calvo-Zaragoza and A. Pertusa and J. Oncina},
issn = {1432-1769},
year = {2017},
date = {2017-01-01},
urldate = {2017-01-01},
journal = {Machine Vision and Applications},
pages = {1-10},
abstract = {Staff-line removal is an important preprocessing stage for most optical music recognition systems. Common procedures to solve this task involve image processing techniques. In contrast to these traditional methods based on hand-engineered transformations, the problem can also be approached as a classification task in which each pixel is labeled as either staff or symbol, so that only those that belong to symbols are kept in the image. In order to perform this classification, we propose the use of convolutional neural networks, which have demonstrated an outstanding performance in image retrieval tasks. The initial features of each pixel consist of a square patch from the input image centered at that pixel. The proposed network is trained by using a dataset which contains pairs of scores with and without the staff lines. Our results in both binary and grayscale images show that the proposed technique is very accurate, outperforming both other classifiers and the state-of-the-art strategies considered. In addition, several advantages of the presented methodology with respect to traditional procedures proposed so far are discussed.},
keywords = {TIMuL},
pubstate = {published},
tppubtype = {article}
}
Staff-line removal is an important preprocessing stage for most optical music recognition systems. Common procedures to solve this task involve image processing techniques. In contrast to these traditional methods based on hand-engineered transformations, the problem can also be approached as a classification task in which each pixel is labeled as either staff or symbol, so that only those that belong to symbols are kept in the image. In order to perform this classification, we propose the use of convolutional neural networks, which have demonstrated an outstanding performance in image retrieval tasks. The initial features of each pixel consist of a square patch from the input image centered at that pixel. The proposed network is trained by using a dataset which contains pairs of scores with and without the staff lines. Our results in both binary and grayscale images show that the proposed technique is very accurate, outperforming both other classifiers and the state-of-the-art strategies considered. In addition, several advantages of the presented methodology with respect to traditional procedures proposed so far are discussed. Gallego, A. J.; Calvo-Zaragoza, J.
Staff-line removal with Selectional Auto-Encoders Journal Article
In: Expert Systems With Applications, vol. 89, pp. 138 - 148, 2017, ISSN: 0957-4174.
Abstract | Links | BibTeX | Tags: TIMuL
@article{k372,
title = {Staff-line removal with Selectional Auto-Encoders},
author = {A. J. Gallego and J. Calvo-Zaragoza},
url = {https://grfia.dlsi.ua.es/repositori/grfia/pubs/372/staff-line-removal.pdf},
issn = {0957-4174},
year = {2017},
date = {2017-01-01},
urldate = {2017-01-01},
journal = {Expert Systems With Applications},
volume = {89},
pages = {138 - 148},
abstract = {Staff-line removal is an important preprocessing stage as regards most Optical Music Recognition systems. The common procedures employed to carry out this task involve image processing techniques. In contrast to these traditional methods, which are based on hand-engineered transformations, the problem can also be approached from a machine learning point of view if representative examples of the task are provided. We propose doing this through the use of a new approach involving auto-encoders, which select the appropriate features of an input feature set (Selectional Auto-Encoders). Within the context of the problem at hand, the model is trained to select those pixels of a given image that belong to a musical symbol, thus removing the lines of the staves. Our results show that the proposed technique is quite competitive and significantly outperforms the other state-of-art strategies considered, particularly when dealing with grayscale input images.},
keywords = {TIMuL},
pubstate = {published},
tppubtype = {article}
}
Staff-line removal is an important preprocessing stage as regards most Optical Music Recognition systems. The common procedures employed to carry out this task involve image processing techniques. In contrast to these traditional methods, which are based on hand-engineered transformations, the problem can also be approached from a machine learning point of view if representative examples of the task are provided. We propose doing this through the use of a new approach involving auto-encoders, which select the appropriate features of an input feature set (Selectional Auto-Encoders). Within the context of the problem at hand, the model is trained to select those pixels of a given image that belong to a musical symbol, thus removing the lines of the staves. Our results show that the proposed technique is quite competitive and significantly outperforms the other state-of-art strategies considered, particularly when dealing with grayscale input images.2016
Rizo, D.; Calvo-Zaragoza, J.; Iñesta, J. M.; Illescas, P. R.
Hidden Markov Models for Functional Analysis Proceedings Article
In: Music and Machine Learning Workshop, Riva del Garda, 2016.
@inproceedings{k370,
title = {Hidden Markov Models for Functional Analysis},
author = {D. Rizo and J. Calvo-Zaragoza and J. M. Iñesta and P. R. Illescas},
year = {2016},
date = {2016-09-01},
booktitle = {Music and Machine Learning Workshop, Riva del Garda},
keywords = {TIMuL},
pubstate = {published},
tppubtype = {inproceedings}
}
Valero-Mas, J. J.; Calvo-Zaragoza, J.; Rico-Juan, J. R.
On the suitability of Prototype Selection methods for kNN classification with distributed data Journal Article
In: Neurocomputing, vol. 203, pp. 150-160, 2016.
@article{k341,
title = {On the suitability of Prototype Selection methods for kNN classification with distributed data},
author = {J. J. Valero-Mas and J. Calvo-Zaragoza and J. R. Rico-Juan},
url = {https://grfia.dlsi.ua.es/repositori/grfia/pubs/341/SuitabilityPSDistributedScenarios.pdf},
year = {2016},
date = {2016-08-01},
journal = {Neurocomputing},
volume = {203},
pages = {150-160},
keywords = {TIMuL},
pubstate = {published},
tppubtype = {article}
}
Calvo-Zaragoza, J.; Rizo, D.; Iñesta, J. M.
Two (note) heads are better than one: pen-based multimodal interaction with music scores Proceedings Article
In: Devaney, J. (Ed.): 17th International Society for Music Information Retrieval Conference, pp. 509-514, New York City, 2016, ISBN: 978-0-692-75506-8.
@inproceedings{k345,
title = {Two (note) heads are better than one: pen-based multimodal interaction with music scores},
author = {J. Calvo-Zaragoza and D. Rizo and J. M. Iñesta},
editor = {J. Devaney},
url = {https://grfia.dlsi.ua.es/repositori/grfia/pubs/345/two-note-heads.pdf},
isbn = {978-0-692-75506-8},
year = {2016},
date = {2016-08-01},
booktitle = {17th International Society for Music Information Retrieval Conference},
pages = {509-514},
address = {New York City},
keywords = {TIMuL},
pubstate = {published},
tppubtype = {inproceedings}
}
Calvo-Zaragoza, J.; Oncina, J.; Higuera, C. De La
Computing the Expected Edit Distance from a String to a PFA Proceedings Article
In: Han, Yo-Sub; Salomaa, Kai (Ed.): 21st International Conference Implementation and Application of Automata, pp. 39-50, Springer, 2016.
@inproceedings{k342,
title = {Computing the Expected Edit Distance from a String to a PFA},
author = {J. Calvo-Zaragoza and J. Oncina and C. De La Higuera},
editor = {Yo-Sub Han and Kai Salomaa},
url = {https://grfia.dlsi.ua.es/repositori/grfia/pubs/342/distance-string-pfa.pdf},
year = {2016},
date = {2016-07-01},
urldate = {2016-07-01},
booktitle = {21st International Conference Implementation and Application of Automata},
pages = {39-50},
publisher = {Springer},
keywords = {TIMuL},
pubstate = {published},
tppubtype = {inproceedings}
}
Calvo-Zaragoza, J.; Micó, L.; Oncina, J.
Music staff removal with supervised pixel classification Journal Article
In: International Journal on Document Analysis and Recognition, vol. 19, no. 3, pp. 211-219, 2016, ISSN: 1433-2833.
@article{k336,
title = {Music staff removal with supervised pixel classification},
author = {J. Calvo-Zaragoza and L. Micó and J. Oncina},
url = {https://grfia.dlsi.ua.es/repositori/grfia/pubs/336/classification-approach-staff.pdf},
issn = {1433-2833},
year = {2016},
date = {2016-01-01},
journal = {International Journal on Document Analysis and Recognition},
volume = {19},
number = {3},
pages = {211-219},
keywords = {TIMuL},
pubstate = {published},
tppubtype = {article}
}
Calvo-Zaragoza, J.; Valero-Mas, J. J.; Rico-Juan, J. R.
Prototype Generation on Structural Data using Dissimilarity Space Representation Journal Article
In: Neural Computing and Applications, 2016.
@article{k337,
title = {Prototype Generation on Structural Data using Dissimilarity Space Representation},
author = {J. Calvo-Zaragoza and J. J. Valero-Mas and J. R. Rico-Juan},
url = {https://grfia.dlsi.ua.es/repositori/grfia/pubs/337/prototype-generation-structural.pdf},
year = {2016},
date = {2016-01-01},
journal = {Neural Computing and Applications},
keywords = {TIMuL},
pubstate = {published},
tppubtype = {article}
}
Calvo-Zaragoza, J.; Valero-Mas, J. J.; Rico-Juan, J. R.
Selecting promising classes from generated data for an efficient multi-class NN classification Journal Article
In: Soft Computing, 2016.
@article{k340,
title = {Selecting promising classes from generated data for an efficient multi-class NN classification},
author = {J. Calvo-Zaragoza and J. J. Valero-Mas and J. R. Rico-Juan},
url = {https://grfia.dlsi.ua.es/repositori/grfia/pubs/340/selecting-promising-classes.pdf},
year = {2016},
date = {2016-01-01},
journal = {Soft Computing},
keywords = {TIMuL},
pubstate = {published},
tppubtype = {article}
}
Calvo-Zaragoza, J.; Toselli, A. H.; Vidal, E.
Early Handwritten Music Recognition with Hidden Markov Models Proceedings Article
In: 15th International Conference on Frontiers in Handwriting Recognition, 2016.
@inproceedings{k350,
title = {Early Handwritten Music Recognition with Hidden Markov Models},
author = {J. Calvo-Zaragoza and A. H. Toselli and E. Vidal},
url = {https://grfia.dlsi.ua.es/repositori/grfia/pubs/350/musicNoteRecogIcfhr16.pdf},
year = {2016},
date = {2016-01-01},
booktitle = {15th International Conference on Frontiers in Handwriting Recognition},
keywords = {TIMuL},
pubstate = {published},
tppubtype = {inproceedings}
}
2020
Ríos-Vila, A.; Calvo-Zaragoza, J.; Iñesta, J. M.
Exploring the two-dimensional nature of music notation for score recognition with end-to-end approaches Proceedings Article
In: Proc. of 17th International Conference on Frontiers in Handwriting Recognition (ICFHR), pp. 193–198, IEEE Computer Society IEEE, Dormund (Germany), 2020, ISBN: 978-1-7281-9966-5.
Abstract | BibTeX | Tags: HispaMus
@inproceedings{k461,
title = {Exploring the two-dimensional nature of music notation for score recognition with end-to-end approaches},
author = {A. Ríos-Vila and J. Calvo-Zaragoza and J. M. Iñesta},
isbn = {978-1-7281-9966-5},
year = {2020},
date = {2020-01-01},
booktitle = {Proc. of 17th International Conference on Frontiers in Handwriting Recognition (ICFHR)},
pages = {193--198},
publisher = {IEEE},
address = {Dormund (Germany)},
organization = {IEEE Computer Society},
abstract = {Optical Music Recognition workflows perform several steps to retrieve the content in music score images, being symbol recognition one of the key stages. State-of-the-art approaches for this stage currently address the coding of the output symbols as if they were plain text characters. However, music symbols have a two-dimensional nature that is ignored in these approaches. In this paper, we explore alternative output representations to perform music symbol recognition with state-of-the-art end-to-end neural technologies. We propose and describe new output representations which take into account the mentioned two-dimensional nature. We seek answers to the question of whether it is possible to obtain better recognition results in both printed and handwritten music scores. In this analysis, we compare the results given using three output encodings and two neural approaches. We found that one of the proposed encodings outperforms the results obtained by the standard one. This permits us to conclude that it is interesting to keep researching on this topic to improve end-to-end music score recognition.},
keywords = {HispaMus},
pubstate = {published},
tppubtype = {inproceedings}
}
Optical Music Recognition workflows perform several steps to retrieve the content in music score images, being symbol recognition one of the key stages. State-of-the-art approaches for this stage currently address the coding of the output symbols as if they were plain text characters. However, music symbols have a two-dimensional nature that is ignored in these approaches. In this paper, we explore alternative output representations to perform music symbol recognition with state-of-the-art end-to-end neural technologies. We propose and describe new output representations which take into account the mentioned two-dimensional nature. We seek answers to the question of whether it is possible to obtain better recognition results in both printed and handwritten music scores. In this analysis, we compare the results given using three output encodings and two neural approaches. We found that one of the proposed encodings outperforms the results obtained by the standard one. This permits us to conclude that it is interesting to keep researching on this topic to improve end-to-end music score recognition.
Gallego, A. J.; Calvo-Zaragoza, J.; Rico-Juan, J. R.
Insights Into Efficient k-Nearest Neighbor Classification With Convolutional Neural Codes Journal Article
In: IEEE Access, vol. 8, pp. 99312-99326, 2020, ISSN: 2169-3536.
Abstract | BibTeX | Tags: HispaMus
@article{k444,
title = {Insights Into Efficient k-Nearest Neighbor Classification With Convolutional Neural Codes},
author = {A. J. Gallego and J. Calvo-Zaragoza and J. R. Rico-Juan},
issn = {2169-3536},
year = {2020},
date = {2020-01-01},
urldate = {2020-01-01},
journal = {IEEE Access},
volume = {8},
pages = {99312-99326},
abstract = {The increasing consideration of Convolutional Neural Networks (CNN) has not prevented the use of the k-Nearest Neighbor (kNN) method. In fact, a hybrid CNN-kNN approach is an interesting option in which the network specializes in feature extraction through its activations (Neural Codes), while the kNN has the advantage of performing a retrieval by means of similarity. However, this hybrid approach also has the disadvantages of the kNN search, and especially its high computational cost which is, in principle, undesirable for large-scale data. In this paper, we present the first comprehensive study of efficient kNN search algorithms using this hybrid CNN-kNN approach. This has been done by considering up to 16 different algorithms, each of which is evaluated with a different parametrization, in 7 datasets of heterogeneous composition. Our results show that no single algorithm is capable of covering all aspects, but rather that each family of algorithms is better suited to specific aspects of the problem. This signifies that Fast Similarity Search algorithms maintain their performance, but do not reduce the cost as much as the Data Reduction family does. In turn, the Approximated Similarity Search family is postulated as a good option when attempting to balance accuracy and efficiency. The experiments also suggest that considering statistical transformation algorithms such as Linear Discriminant Analysis might be useful in certain cases.},
keywords = {HispaMus},
pubstate = {published},
tppubtype = {article}
}
The increasing consideration of Convolutional Neural Networks (CNN) has not prevented the use of the k-Nearest Neighbor (kNN) method. In fact, a hybrid CNN-kNN approach is an interesting option in which the network specializes in feature extraction through its activations (Neural Codes), while the kNN has the advantage of performing a retrieval by means of similarity. However, this hybrid approach also has the disadvantages of the kNN search, and especially its high computational cost which is, in principle, undesirable for large-scale data. In this paper, we present the first comprehensive study of efficient kNN search algorithms using this hybrid CNN-kNN approach. This has been done by considering up to 16 different algorithms, each of which is evaluated with a different parametrization, in 7 datasets of heterogeneous composition. Our results show that no single algorithm is capable of covering all aspects, but rather that each family of algorithms is better suited to specific aspects of the problem. This signifies that Fast Similarity Search algorithms maintain their performance, but do not reduce the cost as much as the Data Reduction family does. In turn, the Approximated Similarity Search family is postulated as a good option when attempting to balance accuracy and efficiency. The experiments also suggest that considering statistical transformation algorithms such as Linear Discriminant Analysis might be useful in certain cases.
Rizo, D.; Calvo-Zaragoza, J.; Iñesta, J. M.
Preservación del patrimonio musical mediante transcripción asistida por ordenador Book Chapter
In: Isusi-Fagoaga, R.; Serrano, F. Villanueva (Ed.): La música de la Corona d’Aragó: investigació, transferència i educació, Chapter 15, pp. 275–298, Universitat de València, 2020, ISBN: 978-84-09-19985-3.
@inbook{k466,
title = {Preservación del patrimonio musical mediante transcripción asistida por ordenador},
author = {D. Rizo and J. Calvo-Zaragoza and J. M. Iñesta},
editor = {R. Isusi-Fagoaga and F. Villanueva Serrano},
isbn = {978-84-09-19985-3},
year = {2020},
date = {2020-01-01},
urldate = {2020-01-01},
booktitle = {La música de la Corona d’Aragó: investigació, transferència i educació},
pages = {275--298},
publisher = {Universitat de València},
chapter = {15},
keywords = {HispaMus},
pubstate = {published},
tppubtype = {inbook}
}
2019
Calvo-Zaragoza, J.; Toselli, A. H.; Vidal, E.
Handwritten Music Recognition for Mensural Notation with Convolutional Recurrent Neural Networks Journal Article
In: Pattern Recognition Letters, vol. 128, pp. 115–121, 2019.
Links | BibTeX | Tags: HispaMus
@article{k415,
title = {Handwritten Music Recognition for Mensural Notation with Convolutional Recurrent Neural Networks},
author = {J. Calvo-Zaragoza and A. H. Toselli and E. Vidal},
url = {https://grfia.dlsi.ua.es/repositori/grfia/pubs/415/1-s2.0-S0167865519302338-main.pdf},
year = {2019},
date = {2019-12-01},
journal = {Pattern Recognition Letters},
volume = {128},
pages = {115--121},
keywords = {HispaMus},
pubstate = {published},
tppubtype = {article}
}
Iñesta, J. M.; Rizo, D.; Calvo-Zaragoza, J.
MuRET as a software for the transcription of historical archives Proceedings Article
In: Proceedings of the 2nd Workshop on Reading Music Systems, WoRMS, pp. 12–15, Delft (The Nederlands), 2019.
Abstract | Links | BibTeX | Tags: HispaMus
@inproceedings{k437,
title = {MuRET as a software for the transcription of historical archives},
author = {J. M. Iñesta and D. Rizo and J. Calvo-Zaragoza},
url = {https://grfia.dlsi.ua.es/repositori/grfia/pubs/437/worms-2019-inesta.pdf},
year = {2019},
date = {2019-11-01},
booktitle = {Proceedings of the 2nd Workshop on Reading Music Systems, WoRMS},
pages = {12--15},
address = {Delft (The Nederlands)},
abstract = {The transcription process from historical hand- written music manuscripts to a structured digital encoding has been traditionally performed following a fully manual workflow. At most, it has received some technological support in particular stages, like optical music recognition (OMR) of the source images, or transcription to modern notation with music edition applications. Currently, there is no mature and stable enough solution for the OMR problem, and the most used music editors do not support early notations, such as the mensural notation. A new tool called MUsic Recognition, Encoding, and Transcription (MuRET) has been developed, which covers all transcription phases, from the manuscript image to the encoded digital content. MuRET is designed as a machine-learning based research tool, allowing different processing approaches to be used, and producing both the expected transcribed contents in standard encodings and data for the study of the transcription process.},
keywords = {HispaMus},
pubstate = {published},
tppubtype = {inproceedings}
}
The transcription process from historical hand- written music manuscripts to a structured digital encoding has been traditionally performed following a fully manual workflow. At most, it has received some technological support in particular stages, like optical music recognition (OMR) of the source images, or transcription to modern notation with music edition applications. Currently, there is no mature and stable enough solution for the OMR problem, and the most used music editors do not support early notations, such as the mensural notation. A new tool called MUsic Recognition, Encoding, and Transcription (MuRET) has been developed, which covers all transcription phases, from the manuscript image to the encoded digital content. MuRET is designed as a machine-learning based research tool, allowing different processing approaches to be used, and producing both the expected transcribed contents in standard encodings and data for the study of the transcription process.
Ríos-Vila, A.; Calvo-Zaragoza, J.; Rizo, D.; Iñesta, J. M.
ReadSco: An Open-Source Web-Based Optical Music Recognition Tool Proceedings Article
In: Calvo-Zaragoza, J.; Pacha, A. (Ed.): Proc. of the 2nd International Workshop on Reading Music Systems, WoRMS 2019, pp. 27-30, Delft (The Netherlands), 2019.
Abstract | Links | BibTeX | Tags: HispaMus
@inproceedings{k439,
title = {ReadSco: An Open-Source Web-Based Optical Music Recognition Tool},
author = {A. Ríos-Vila and J. Calvo-Zaragoza and D. Rizo and J. M. Iñesta},
editor = {J. Calvo-Zaragoza and A. Pacha},
url = {https://grfia.dlsi.ua.es/repositori/grfia/pubs/439/ReadSco_Worms2019.pdf},
year = {2019},
date = {2019-11-01},
urldate = {2019-11-01},
booktitle = {Proc. of the 2nd International Workshop on Reading Music Systems, WoRMS 2019},
pages = {27-30},
address = {Delft (The Netherlands)},
abstract = {We introduce READSCO, an open-source web-based community tool for Optical Music Recognition. READSCO aims to serve as a connection between research results and practical use. We describe the design decisions considered to both favours a rapid integration of new advances from the research field and
facilitate the community’s participation in its development. The project is still in its planning phase, so this work is a good opportunity to present the main idea and get direct feedback from other researchers.},
keywords = {HispaMus},
pubstate = {published},
tppubtype = {inproceedings}
}
We introduce READSCO, an open-source web-based community tool for Optical Music Recognition. READSCO aims to serve as a connection between research results and practical use. We describe the design decisions considered to both favours a rapid integration of new advances from the research field and
facilitate the community’s participation in its development. The project is still in its planning phase, so this work is a good opportunity to present the main idea and get direct feedback from other researchers.
facilitate the community’s participation in its development. The project is still in its planning phase, so this work is a good opportunity to present the main idea and get direct feedback from other researchers.
Iñesta, J. M.; Rizo, D.; Calvo-Zaragoza, J.
Recognition, encoding, and transcription of early mensural handwritten music Proceedings Article
In: procedings of the 13th International Workshop on Graphics Recognition (GREC 2019), pp. 13-14, Sydney (Australia), 2019.
@inproceedings{k416,
title = {Recognition, encoding, and transcription of early mensural handwritten music},
author = {J. M. Iñesta and D. Rizo and J. Calvo-Zaragoza},
year = {2019},
date = {2019-09-01},
booktitle = {procedings of the 13th International Workshop on Graphics Recognition (GREC 2019)},
pages = {13-14},
address = {Sydney (Australia)},
keywords = {HispaMus},
pubstate = {published},
tppubtype = {inproceedings}
}
Alfaro-Contreras, M.; Calvo-Zaragoza, J.; Iñesta, J. M.
Approaching End-to-end Optical Music Recognition for Homophonic Scores Proceedings Article
In: Iberian Conference on Pattern Recognition and Image Analysis, IbPRIA 2019, pp. 147–158, Springer, Madrid, 2019, ISBN: 978-3-030-31321-0.
Links | BibTeX | Tags: HispaMus
@inproceedings{k419,
title = {Approaching End-to-end Optical Music Recognition for Homophonic Scores},
author = {M. Alfaro-Contreras and J. Calvo-Zaragoza and J. M. Iñesta},
url = {https://link.springer.com/chapter/10.1007/978-3-030-31321-0_13},
isbn = {978-3-030-31321-0},
year = {2019},
date = {2019-08-01},
urldate = {2019-08-01},
booktitle = {Iberian Conference on Pattern Recognition and Image Analysis, IbPRIA 2019},
pages = {147--158},
publisher = {Springer},
address = {Madrid},
keywords = {HispaMus},
pubstate = {published},
tppubtype = {inproceedings}
}
Nuñez-Alcover, Alicia; León, P. J. Ponce; Calvo-Zaragoza, J.
Glyph and Position Classification of Music Symbols in Early Music Manuscripts Proceedings Article
In: Morales, A.; Fierrez, J.; Sánchez, J. Salvador; Ribeiro, B. (Ed.): Proc. of the 9th Iberian Conference on Pattern Recognition and Image Analysis, LNCS vol. 11867, pp. 159-168, AERFAI, APRP Springer, Madrid, Spain, 2019, ISBN: 978-3-030-31331-9.
Abstract | Links | BibTeX | Tags: HispaMus
@inproceedings{k420,
title = {Glyph and Position Classification of Music Symbols in Early Music Manuscripts},
author = {Alicia Nuñez-Alcover and P. J. Ponce León and J. Calvo-Zaragoza},
editor = {A. Morales and J. Fierrez and J. Salvador Sánchez and B. Ribeiro},
url = {https://grfia.dlsi.ua.es/repositori/grfia/pubs/420/IbPRIA2019_paper_91-1.pdf},
isbn = {978-3-030-31331-9},
year = {2019},
date = {2019-07-01},
booktitle = {Proc. of the 9th Iberian Conference on Pattern Recognition and Image Analysis, LNCS vol. 11867},
pages = {159-168},
publisher = {Springer},
address = {Madrid, Spain},
organization = {AERFAI, APRP},
abstract = {Optical Music Recognition is a field of research that automates the reading of musical scores so as to transcribe their content into a structured digital format. When dealing with music manuscripts, the traditional workflow establishes separate stages of detection and classification of musical symbols. In the latter, most of the research has focused on detecting musical glyphs, ignoring that the meaning of a musical symbol is defined by two components: its glyph and its position within the staff. In this paper we study how to perform both glyph and position classification of handwritten musical symbols in early music manuscripts written in white Mensural notation, a common notation system used for the most part of the XVI and XVII centuries. We make use of Convolutional Neural Networks as the classification method, and we tested several alternatives such as using independent models for each component, combining label spaces, or using both multi-input and multi-output models. Our results on early music manuscripts provide insights about the effectiveness and efficiency of each approach.},
keywords = {HispaMus},
pubstate = {published},
tppubtype = {inproceedings}
}
Optical Music Recognition is a field of research that automates the reading of musical scores so as to transcribe their content into a structured digital format. When dealing with music manuscripts, the traditional workflow establishes separate stages of detection and classification of musical symbols. In the latter, most of the research has focused on detecting musical glyphs, ignoring that the meaning of a musical symbol is defined by two components: its glyph and its position within the staff. In this paper we study how to perform both glyph and position classification of handwritten musical symbols in early music manuscripts written in white Mensural notation, a common notation system used for the most part of the XVI and XVII centuries. We make use of Convolutional Neural Networks as the classification method, and we tested several alternatives such as using independent models for each component, combining label spaces, or using both multi-input and multi-output models. Our results on early music manuscripts provide insights about the effectiveness and efficiency of each approach.
Mateiu, T. N.; Gallego, A. J.; Calvo-Zaragoza, J.
Domain Adaptation for Handwritten Symbol Recognition: A Case of Study in Old Music Manuscripts Proceedings Article
In: Iberian Conference on Pattern Recognition and Image Analysis, IbPRIA, pp. 135–146, Springer International Publishing, Madrid, Spain, 2019, ISBN: 978-3-030-31321-0.
Abstract | BibTeX | Tags: HispaMus
@inproceedings{k410,
title = {Domain Adaptation for Handwritten Symbol Recognition: A Case of Study in Old Music Manuscripts},
author = {T. N. Mateiu and A. J. Gallego and J. Calvo-Zaragoza},
isbn = {978-3-030-31321-0},
year = {2019},
date = {2019-07-01},
urldate = {2019-07-01},
booktitle = {Iberian Conference on Pattern Recognition and Image Analysis, IbPRIA},
pages = {135--146},
publisher = {Springer International Publishing},
address = {Madrid, Spain},
abstract = {The existence of a large amount of untranscripted music manuscripts has caused initiatives that use Machine Learning (ML) for Optical Music Recognition, in order to efficiently transcribe the music sources into a machine-readable format. Although most music manuscript are similar in nature, they inevitably vary from one another. This fact can negatively influence the complexity of the classification task because most ML models fail to transfer their knowledge from one domain to another, thereby requiring learning from scratch on new domains after manually labeling new data. This work studies the ability of a Domain Adversarial Neural Network for domain adaptation in the context of classifying handwritten music symbols. The main idea is to exploit the knowledge of a specific manuscript to classify symbols from different (unlabeled) manuscripts. The reported results are promising, obtaining a substantial improvement over a conventional Convolutional Neural Network approach, which can be used as a basis for future research.},
keywords = {HispaMus},
pubstate = {published},
tppubtype = {inproceedings}
}
The existence of a large amount of untranscripted music manuscripts has caused initiatives that use Machine Learning (ML) for Optical Music Recognition, in order to efficiently transcribe the music sources into a machine-readable format. Although most music manuscript are similar in nature, they inevitably vary from one another. This fact can negatively influence the complexity of the classification task because most ML models fail to transfer their knowledge from one domain to another, thereby requiring learning from scratch on new domains after manually labeling new data. This work studies the ability of a Domain Adversarial Neural Network for domain adaptation in the context of classifying handwritten music symbols. The main idea is to exploit the knowledge of a specific manuscript to classify symbols from different (unlabeled) manuscripts. The reported results are promising, obtaining a substantial improvement over a conventional Convolutional Neural Network approach, which can be used as a basis for future research.
Calvo-Zaragoza, J.; Rico-Juan, J. R.; Gallego, A. J.
Ensemble classification from deep predictions with test data augmentation Journal Article
In: Soft Computing, 2019, ISSN: 1433-7479.
@article{k409,
title = {Ensemble classification from deep predictions with test data augmentation},
author = {J. Calvo-Zaragoza and J. R. Rico-Juan and A. J. Gallego},
issn = {1433-7479},
year = {2019},
date = {2019-04-01},
urldate = {2019-04-01},
journal = {Soft Computing},
abstract = {Data augmentation has become a standard step to improve the predictive power and robustness of convolutional neural networks by means of the synthetic generation of new samples depicting different deformations. This step has been traditionally considered to improve the network at the training stage. In this work, however, we study the use of data augmentation at classification time. That is, the test sample is augmented, following the same procedure considered for training, and the decision is taken with an ensemble prediction over all these samples. We present comprehensive experimentation with several datasets and ensemble decisions, considering a rather generic data augmentation procedure. Our results show that performing this step is able to boost the original classification, even when the room for improvement is limited.},
keywords = {},
pubstate = {published},
tppubtype = {article}
}
Data augmentation has become a standard step to improve the predictive power and robustness of convolutional neural networks by means of the synthetic generation of new samples depicting different deformations. This step has been traditionally considered to improve the network at the training stage. In this work, however, we study the use of data augmentation at classification time. That is, the test sample is augmented, following the same procedure considered for training, and the decision is taken with an ensemble prediction over all these samples. We present comprehensive experimentation with several datasets and ensemble decisions, considering a rather generic data augmentation procedure. Our results show that performing this step is able to boost the original classification, even when the room for improvement is limited.
Calvo-Zaragoza, J.; Gallego, A. J.
A selectional auto-encoder approach for document image binarization Journal Article
In: Pattern Recognition, vol. 86, pp. 37-47, 2019, ISSN: 0031-3203.
Abstract | Links | BibTeX | Tags: GRE16-14
@article{k395,
title = {A selectional auto-encoder approach for document image binarization},
author = {J. Calvo-Zaragoza and A. J. Gallego},
url = {https://grfia.dlsi.ua.es/repositori/grfia/pubs/395/1706.10241.pdf},
issn = {0031-3203},
year = {2019},
date = {2019-01-01},
journal = {Pattern Recognition},
volume = {86},
pages = {37-47},
abstract = {Binarization plays a key role in the automatic information retrieval from document images. This process is usually performed in the first stages of document analysis systems, and serves as a basis for subsequent steps. Hence it has to be robust in order to allow the full analysis workflow to be successful. Several methods for document image binarization have been proposed so far, most of which are based on hand-crafted image processing strategies. Recently, Convolutional Neural Networks have shown an amazing performance in many disparate duties related to computer vision. In this paper we discuss the use of convolutional auto-encoders devoted to learning an end-to-end map from an input image to its selectional output, in which activations indicate the likelihood of pixels to be either foreground or background. Once trained, documents can therefore be binarized by parsing them through the model and applying a global threshold. This approach has proven to outperform existing binarization strategies in a number of document types.},
keywords = {GRE16-14},
pubstate = {published},
tppubtype = {article}
}
Binarization plays a key role in the automatic information retrieval from document images. This process is usually performed in the first stages of document analysis systems, and serves as a basis for subsequent steps. Hence it has to be robust in order to allow the full analysis workflow to be successful. Several methods for document image binarization have been proposed so far, most of which are based on hand-crafted image processing strategies. Recently, Convolutional Neural Networks have shown an amazing performance in many disparate duties related to computer vision. In this paper we discuss the use of convolutional auto-encoders devoted to learning an end-to-end map from an input image to its selectional output, in which activations indicate the likelihood of pixels to be either foreground or background. Once trained, documents can therefore be binarized by parsing them through the model and applying a global threshold. This approach has proven to outperform existing binarization strategies in a number of document types.
Rico-Juan, J. R.; Gallego, A. J.; Calvo-Zaragoza, J.
Automatic detection of inconsistencies between numerical scores and textual feedback in peer-assessment processes with machine learning Journal Article
In: Computers and Education, vol. 140, pp. 103609, 2019.
@article{k414,
title = {Automatic detection of inconsistencies between numerical scores and textual feedback in peer-assessment processes with machine learning},
author = {J. R. Rico-Juan and A. J. Gallego and J. Calvo-Zaragoza},
year = {2019},
date = {2019-01-01},
journal = {Computers and Education},
volume = {140},
pages = {103609},
abstract = {The use of peer assessment for open-ended activities has advantages for both teachers and students. Teachers might reduce the workload of the correction process and students achieve a better understanding of the subject by evaluating the activities of their peers. In order to ease the process, it is advisable to provide the students with a rubric over which performing the assessment of their peers and however, restricting themselves to provide only numerical scores is detrimental, as it prevents providing valuable feedback to others peers. Since this assessment produces two modalities of the same evaluation, namely numerical score and textual feedback, it is possible to apply automatic techniques to detect inconsistencies in the evaluation, thus minimizing the teachers' workload for supervising the whole process. This paper proposes a machine learning approach for the detection of such inconsistencies. To this end, we consider two different approaches, each of which is tested with different algorithms, in order to both evaluate the approach itself and find appropriate models to make it successful. The experiments carried out with 4 groups of students and 2 types of activities show that the proposed approach is able to yield reliable results, thus representing a valuable approach for ensuring a fair operation of the peer assessment process.},
keywords = {},
pubstate = {published},
tppubtype = {article}
}
The use of peer assessment for open-ended activities has advantages for both teachers and students. Teachers might reduce the workload of the correction process and students achieve a better understanding of the subject by evaluating the activities of their peers. In order to ease the process, it is advisable to provide the students with a rubric over which performing the assessment of their peers and however, restricting themselves to provide only numerical scores is detrimental, as it prevents providing valuable feedback to others peers. Since this assessment produces two modalities of the same evaluation, namely numerical score and textual feedback, it is possible to apply automatic techniques to detect inconsistencies in the evaluation, thus minimizing the teachers' workload for supervising the whole process. This paper proposes a machine learning approach for the detection of such inconsistencies. To this end, we consider two different approaches, each of which is tested with different algorithms, in order to both evaluate the approach itself and find appropriate models to make it successful. The experiments carried out with 4 groups of students and 2 types of activities show that the proposed approach is able to yield reliable results, thus representing a valuable approach for ensuring a fair operation of the peer assessment process.
Rico-Juan, J. R.; Valero-Mas, J. J.; Calvo-Zaragoza, J.
Extensions to rank-based prototype selection in k-Nearest Neighbour classification Journal Article
In: Applied Soft Computing, no. 105803, 2019, ISSN: 1568-4946.
@article{k418,
title = {Extensions to rank-based prototype selection in k-Nearest Neighbour classification},
author = {J. R. Rico-Juan and J. J. Valero-Mas and J. Calvo-Zaragoza},
issn = {1568-4946},
year = {2019},
date = {2019-01-01},
journal = {Applied Soft Computing},
number = {105803},
abstract = {The k-nearest neighbour rule is commonly considered for classification tasks given its straightforward implementation and good performance in many applications. However, its efficiency represents an obstacle in real-case scenarios because the classification requires computing a distance to every single prototype of the training set. Prototype Selection (PS) is a typical approach to alleviate this problem, which focuses on reducing the size of the training set by selecting the most interesting prototypes. In this context, rank methods have been postulated as a good solution: following some heuristics, these methods perform an ordering of the prototypes according to their relevance in the classification task, which is then used to select the most relevant ones. This work presents a significant improvement of existing rank methods by proposing two extensions: i) a greater robustness against noise at label level by considering the parameter `k' of the classification in the selection process and and ii) a new parameter-free rule to select the prototypes once they have been ordered. The experiments performed in different scenarios and datasets demonstrate the goodness of these extensions. Also, it is reported that the new full approach is competitive with respect to existing PS algorithms.},
keywords = {},
pubstate = {published},
tppubtype = {article}
}
The k-nearest neighbour rule is commonly considered for classification tasks given its straightforward implementation and good performance in many applications. However, its efficiency represents an obstacle in real-case scenarios because the classification requires computing a distance to every single prototype of the training set. Prototype Selection (PS) is a typical approach to alleviate this problem, which focuses on reducing the size of the training set by selecting the most interesting prototypes. In this context, rank methods have been postulated as a good solution: following some heuristics, these methods perform an ordering of the prototypes according to their relevance in the classification task, which is then used to select the most relevant ones. This work presents a significant improvement of existing rank methods by proposing two extensions: i) a greater robustness against noise at label level by considering the parameter `k' of the classification in the selection process and and ii) a new parameter-free rule to select the prototypes once they have been ordered. The experiments performed in different scenarios and datasets demonstrate the goodness of these extensions. Also, it is reported that the new full approach is competitive with respect to existing PS algorithms.
Baró, A.; Riba, P.; Calvo-Zaragoza, J.; Fornés, A.
From Optical Music Recognition to Handwritten Music Recognition: A baseline Journal Article
In: Pattern Recognition Letters, vol. 123, pp. 1-8, 2019.
BibTeX | Tags:
@article{k428,
title = {From Optical Music Recognition to Handwritten Music Recognition: A baseline},
author = {A. Baró and P. Riba and J. Calvo-Zaragoza and A. Fornés},
year = {2019},
date = {2019-01-01},
journal = {Pattern Recognition Letters},
volume = {123},
pages = {1-8},
keywords = {},
pubstate = {published},
tppubtype = {article}
}
Román, M. A.; Pertusa, A.; Calvo-Zaragoza, J.
A Holistic Approach to Polyphonic Music Transcription with Neural Networks Proceedings Article
In: Flexer, Julián Urbano Geoffroy Peeters Arthur (Ed.): Proc. of the 20th International Society for Music Information Retrieval Conference, ISMIR, pp. 731–737, Delf, The Netherlands, 2019.
@inproceedings{k442,
title = {A Holistic Approach to Polyphonic Music Transcription with Neural Networks},
author = {M. A. Román and A. Pertusa and J. Calvo-Zaragoza},
editor = {Julián Urbano Geoffroy Peeters Arthur Flexer},
year = {2019},
date = {2019-01-01},
booktitle = {Proc. of the 20th International Society for Music Information Retrieval Conference, ISMIR},
pages = {731--737},
address = {Delf, The Netherlands},
keywords = {HispaMus},
pubstate = {published},
tppubtype = {inproceedings}
}
Garay-Maestre, U.; Gallego, A. J.; Calvo-Zaragoza, J.
Data Augmentation via Variational Auto-Encoders Proceedings Article
In: Progress in Pattern Recognition, Image Analysis, Computer Vision, and Applications, pp. 29–37, 2019, ISBN: 978-3-030-13469-3.
Abstract | BibTeX | Tags: HispaMus
@inproceedings{k406,
title = {Data Augmentation via Variational Auto-Encoders},
author = {U. Garay-Maestre and A. J. Gallego and J. Calvo-Zaragoza},
isbn = {978-3-030-13469-3},
year = {2019},
date = {2019-01-01},
urldate = {2019-01-01},
booktitle = {Progress in Pattern Recognition, Image Analysis, Computer Vision, and Applications},
pages = {29--37},
abstract = {Data augmentation is a widely considered technique to improve the performance of Convolutional Neural Networks during training. This step consists in synthetically generate new labeled data by perturbing the samples of the training set, which is expected to provide more robustness to the learning process. The problem is that the augmentation procedure has to be adjusted manually because the perturbations considered must make sense for the task at issue. In this paper we propose the use of Variational Auto-Encoders (VAEs) to generate new synthetic samples, instead of resorting to heuristic strategies. VAEs are powerful generative models that learn a parametric latent space of the input domain from which new samples can be generated. In our experiments over the well-known MNIST dataset, the data augmentation by VAEs improves the base results, yet to a lesser extent of that obtained by a well-adjusted conventional data augmentation. However, the combination of both conventional and VAE-guided data augmentations outperforms all the results, thereby demonstrating the goodness of our proposal.},
keywords = {HispaMus},
pubstate = {published},
tppubtype = {inproceedings}
}
Data augmentation is a widely considered technique to improve the performance of Convolutional Neural Networks during training. This step consists in synthetically generate new labeled data by perturbing the samples of the training set, which is expected to provide more robustness to the learning process. The problem is that the augmentation procedure has to be adjusted manually because the perturbations considered must make sense for the task at issue. In this paper we propose the use of Variational Auto-Encoders (VAEs) to generate new synthetic samples, instead of resorting to heuristic strategies. VAEs are powerful generative models that learn a parametric latent space of the input domain from which new samples can be generated. In our experiments over the well-known MNIST dataset, the data augmentation by VAEs improves the base results, yet to a lesser extent of that obtained by a well-adjusted conventional data augmentation. However, the combination of both conventional and VAE-guided data augmentations outperforms all the results, thereby demonstrating the goodness of our proposal.
2018
Rico-Juan, J. R.; Gallego, A. J.; Valero-Mas, J. J.; Calvo-Zaragoza, J.
Statistical semi-supervised system for grading multiple peer-reviewed open-ended works Journal Article
In: Computers & Education, vol. 126, no. 1, pp. 264-282, 2018.
Abstract | Links | BibTeX | Tags:
@article{k394,
title = {Statistical semi-supervised system for grading multiple peer-reviewed open-ended works},
author = {J. R. Rico-Juan and A. J. Gallego and J. J. Valero-Mas and J. Calvo-Zaragoza},
url = {https://grfia.dlsi.ua.es/repositori/grfia/pubs/394/statistical-semi-supervised.pdf},
year = {2018},
date = {2018-11-01},
journal = {Computers & Education},
volume = {126},
number = {1},
pages = {264-282},
abstract = {In the education context, open-ended works generally entail a series of benefits as the possibility of develop original ideas and a more productive learning process to the student rather than closed-answer activities. Nevertheless, such works suppose a significant correction workload to the teacher in contrast to the latter ones that can be self-corrected. Furthermore, such workload turns to be intractable with large groups of students. In order to maintain the advantages of open-ended works with a reasonable amount of correction effort, this article proposes a novel methodology: students perform the corrections using a rubric (closed Likert scale) as a guideline in a peer-review fashion and then, their markings are automatically analyzed with statistical tools to detect possible biased scorings and finally, in the event the statistical analysis detects a biased case, the teacher is required to intervene to manually correct the assignment. This methodology has been tested on two different assignments with two heterogeneous groups of people to assess the robustness and reliability of the proposal. As a result, we obtain values over 95% in the confidence of the intra-class correlation test (ICC) between the grades computed by our proposal and those directly resulting from the manual correction of the teacher. These figures confirm that the evaluation obtained with the proposed methodology is statistically similar to that of the manual correction of the teacher with a remarkable decrease in terms of effort.},
keywords = {},
pubstate = {published},
tppubtype = {article}
}
In the education context, open-ended works generally entail a series of benefits as the possibility of develop original ideas and a more productive learning process to the student rather than closed-answer activities. Nevertheless, such works suppose a significant correction workload to the teacher in contrast to the latter ones that can be self-corrected. Furthermore, such workload turns to be intractable with large groups of students. In order to maintain the advantages of open-ended works with a reasonable amount of correction effort, this article proposes a novel methodology: students perform the corrections using a rubric (closed Likert scale) as a guideline in a peer-review fashion and then, their markings are automatically analyzed with statistical tools to detect possible biased scorings and finally, in the event the statistical analysis detects a biased case, the teacher is required to intervene to manually correct the assignment. This methodology has been tested on two different assignments with two heterogeneous groups of people to assess the robustness and reliability of the proposal. As a result, we obtain values over 95% in the confidence of the intra-class correlation test (ICC) between the grades computed by our proposal and those directly resulting from the manual correction of the teacher. These figures confirm that the evaluation obtained with the proposed methodology is statistically similar to that of the manual correction of the teacher with a remarkable decrease in terms of effort.
Rizo, D.; Calvo-Zaragoza, J.; Iñesta, J. M.
MuRET: a music recognition, encoding, and transcription tool Proceedings Article
In: Proceedings of the 5th International Conference on Digital Libraries for Musicology (DLfM'18), pp. 52–56, ACM, París, France, 2018, ISBN: 978-1-4503-6522-2.
Abstract | BibTeX | Tags: HispaMus
@inproceedings{k397,
title = {MuRET: a music recognition, encoding, and transcription tool},
author = {D. Rizo and J. Calvo-Zaragoza and J. M. Iñesta},
isbn = {978-1-4503-6522-2},
year = {2018},
date = {2018-09-01},
booktitle = {Proceedings of the 5th International Conference on Digital Libraries for Musicology (DLfM'18)},
pages = {52--56},
publisher = {ACM},
address = {París, France},
abstract = {The transcription process from early and modern notation ma-nu-scripts to a structured digital encoding has been traditionally performed following a fully manual workflow. At most it has received some technological support in particular stages, like optical music recognition (OMR) of the source images, or transcription to modern notation with music edition applications. Currently, there is no mature and stable enough solution for the OMR problem, and the most used music editors do not support early notations, such as the mensural one. In this work, a new tool called MUsic Recognition, Encoding, and Transcription (MuRET) is introduced, which covers all transcription phases, from the manuscript source to the encoded digital content. MuRET is designed as a technology-focused research tool, allowing different processing approaches to be used, and producing both the expected transcribed contents in standard encodings and data for the study of the transcription process itself.},
keywords = {HispaMus},
pubstate = {published},
tppubtype = {inproceedings}
}
The transcription process from early and modern notation ma-nu-scripts to a structured digital encoding has been traditionally performed following a fully manual workflow. At most it has received some technological support in particular stages, like optical music recognition (OMR) of the source images, or transcription to modern notation with music edition applications. Currently, there is no mature and stable enough solution for the OMR problem, and the most used music editors do not support early notations, such as the mensural one. In this work, a new tool called MUsic Recognition, Encoding, and Transcription (MuRET) is introduced, which covers all transcription phases, from the manuscript source to the encoded digital content. MuRET is designed as a technology-focused research tool, allowing different processing approaches to be used, and producing both the expected transcribed contents in standard encodings and data for the study of the transcription process itself.
Castellanos, F. J.; Calvo-Zaragoza, J.; Vigliensoni, G.; Fujinaga, I.
Document Analysis of Music Score Images with Selectional Auto-Encoders Proceedings Article
In: Proceedings of the19th International Society for Music Information Retrieval Conference, pp. 256-263, ISMIR 2018 Paris, France, 2018.
@inproceedings{k449,
title = {Document Analysis of Music Score Images with Selectional Auto-Encoders},
author = {F. J. Castellanos and J. Calvo-Zaragoza and G. Vigliensoni and I. Fujinaga},
year = {2018},
date = {2018-09-01},
booktitle = {Proceedings of the19th International Society for Music Information Retrieval Conference},
pages = {256-263},
address = {Paris, France},
organization = {ISMIR 2018},
keywords = {HispaMus},
pubstate = {published},
tppubtype = {inproceedings}
}
Calvo-Zaragoza, J.; Rizo, D.
Camera-PrIMuS: Neural End-to-End Optical Music Recognition on Realistic Monophonic Scores Proceedings Article
In: Proceedings of the 19th International Society of Music Information Retrieval (ISMIR), International Society of Music Information Retrieval 2018.
Links | BibTeX | Tags: HispaMus
@inproceedings{k391,
title = {Camera-PrIMuS: Neural End-to-End Optical Music Recognition on Realistic Monophonic Scores},
author = {J. Calvo-Zaragoza and D. Rizo},
url = {https://grfia.dlsi.ua.es/repositori/grfia/pubs/391/33_Paper.pdf},
year = {2018},
date = {2018-09-01},
urldate = {2018-09-01},
booktitle = {Proceedings of the 19th International Society of Music Information Retrieval (ISMIR)},
organization = {International Society of Music Information Retrieval},
keywords = {HispaMus},
pubstate = {published},
tppubtype = {inproceedings}
}
Román, M. A.; Pertusa, A.; Calvo-Zaragoza, J.
An End-to-End Framework for Audio-to-Score Music Transcription on Monophonic Excerpts Proceedings Article
In: Proc. of the 19th International Society for Music Information Retrieval Conference (ISMIR), Paris, France, 2018.
BibTeX | Tags: GRE16-14, HispaMus
@inproceedings{k389,
title = {An End-to-End Framework for Audio-to-Score Music Transcription on Monophonic Excerpts},
author = {M. A. Román and A. Pertusa and J. Calvo-Zaragoza},
year = {2018},
date = {2018-09-01},
urldate = {2018-09-01},
booktitle = {Proc. of the 19th International Society for Music Information Retrieval Conference (ISMIR)},
address = {Paris, France},
keywords = {GRE16-14, HispaMus},
pubstate = {published},
tppubtype = {inproceedings}
}
Iñesta, J. M.; León, P. J. Ponce; Rizo, D.; Oncina, J.; Micó, L.; Rico-Juan, J. R.; Pérez-Sancho, C.; Pertusa, A.
HISPAMUS: Handwritten Spanish Music Heritage Preservation by Automatic Transcription Proceedings Article
In: Calvo-Zaragoza, J.; Jr., J. Hajic; Pacha, A. (Ed.): Proceedings of the 1st Workshop on Reading Music Systems, WoRMS, pp. 17–18, Paris, 2018.
Abstract | Links | BibTeX | Tags: HispaMus
@inproceedings{k396,
title = {HISPAMUS: Handwritten Spanish Music Heritage Preservation by Automatic Transcription},
author = {J. M. Iñesta and P. J. Ponce León and D. Rizo and J. Oncina and L. Micó and J. R. Rico-Juan and C. Pérez-Sancho and A. Pertusa},
editor = {J. Calvo-Zaragoza and J. Hajic Jr. and A. Pacha},
url = {https://grfia.dlsi.ua.es/repositori/grfia/pubs/396/worms-2018-inesta.pdf},
year = {2018},
date = {2018-09-01},
urldate = {2018-09-01},
booktitle = {Proceedings of the 1st Workshop on Reading Music Systems, WoRMS},
pages = {17--18},
address = {Paris},
abstract = {The HISPAMUS proposal aims at enhancing the Hispanic music heritage from the 15th to the 19th centuries, by exploiting the digital resources of these collections. In addition, thousands of oral tradition melodies that were compiled by folklorists in the 1950s decade are digitized just as images, currently without the possibility of content-based search or study. It is necessary to develop services and tools for the benefit of archives, libraries, scholars, computer scientists and general public. HISPAMUS tries to provide smart access to archival manuscripts of music scores, allowing its reuse and exploitation. In order to reach this ambitious goal, our group can provide cutting-edge technology in the fields of Machine Learning, Pattern Recognition, and Optical Music Recognition.},
keywords = {HispaMus},
pubstate = {published},
tppubtype = {inproceedings}
}
The HISPAMUS proposal aims at enhancing the Hispanic music heritage from the 15th to the 19th centuries, by exploiting the digital resources of these collections. In addition, thousands of oral tradition melodies that were compiled by folklorists in the 1950s decade are digitized just as images, currently without the possibility of content-based search or study. It is necessary to develop services and tools for the benefit of archives, libraries, scholars, computer scientists and general public. HISPAMUS tries to provide smart access to archival manuscripts of music scores, allowing its reuse and exploitation. In order to reach this ambitious goal, our group can provide cutting-edge technology in the fields of Machine Learning, Pattern Recognition, and Optical Music Recognition.
Calvo-Zaragoza, J.; Rizo, D.
End-to-End Neural Optical Music Recognition of Monophonic Scores Journal Article
In: Applied Sciences, vol. 8, no. 4, pp. 606–623, 2018, ISSN: 2076-3417.
@article{k390,
title = {End-to-End Neural Optical Music Recognition of Monophonic Scores},
author = {J. Calvo-Zaragoza and D. Rizo},
issn = {2076-3417},
year = {2018},
date = {2018-04-01},
journal = {Applied Sciences},
volume = {8},
number = {4},
pages = {606--623},
keywords = {HispaMus},
pubstate = {published},
tppubtype = {article}
}
Castellanos, F. J.; Valero-Mas, J. J.; Calvo-Zaragoza, J.; Rico-Juan, J. R.
Oversampling imbalanced data in the string space Journal Article
In: Pattern Recognition Letters, vol. 103, pp. 32–38, 2018, ISSN: 0167-8655.
Abstract | BibTeX | Tags: GRE16-14
@article{k382,
title = {Oversampling imbalanced data in the string space},
author = {F. J. Castellanos and J. J. Valero-Mas and J. Calvo-Zaragoza and J. R. Rico-Juan},
issn = {0167-8655},
year = {2018},
date = {2018-02-01},
journal = {Pattern Recognition Letters},
volume = {103},
pages = {32--38},
abstract = {Imbalanced data is a typical problem in the supervised classification field, which occurs when the different classes are not equally represented. This fact typically results in the classifier biasing its performance towards the class representing the majority of the elements. Many methods have been proposed to alleviate this scenario, yet all of them assume that data is represented as feature vectors. In this paper we propose a strategy to balance a dataset whose samples are encoded as strings. Our approach is based on adapting the well-known Synthetic Minority Over-sampling Technique (SMOTE) algorithm to the string space. More precisely, data generation is achieved with an iterative approach to create artificial strings within the segment between two given samples of the training set. Results with several datasets and imbalance ratios show that the proposed strategy properly deals with the problem in all cases considered.},
keywords = {GRE16-14},
pubstate = {published},
tppubtype = {article}
}
Imbalanced data is a typical problem in the supervised classification field, which occurs when the different classes are not equally represented. This fact typically results in the classifier biasing its performance towards the class representing the majority of the elements. Many methods have been proposed to alleviate this scenario, yet all of them assume that data is represented as feature vectors. In this paper we propose a strategy to balance a dataset whose samples are encoded as strings. Our approach is based on adapting the well-known Synthetic Minority Over-sampling Technique (SMOTE) algorithm to the string space. More precisely, data generation is achieved with an iterative approach to create artificial strings within the segment between two given samples of the training set. Results with several datasets and imbalance ratios show that the proposed strategy properly deals with the problem in all cases considered.
Gallego, A. J.; Pertusa, A.; Calvo-Zaragoza, J.
Improving Convolutional Neural Networks’ Accuracy in Noisy Environments Using k-Nearest Neighbors Journal Article
In: Applied Sciences, vol. 8, no. 11, 2018, ISSN: 2076-3417.
Abstract | BibTeX | Tags: HispaMus
@article{k399,
title = {Improving Convolutional Neural Networks’ Accuracy in Noisy Environments Using k-Nearest Neighbors},
author = {A. J. Gallego and A. Pertusa and J. Calvo-Zaragoza},
issn = {2076-3417},
year = {2018},
date = {2018-01-01},
journal = {Applied Sciences},
volume = {8},
number = {11},
abstract = {We present a hybrid approach to improve the accuracy of Convolutional Neural Networks (CNN) without retraining the model. The proposed architecture replaces the softmax layer by a k-Nearest Neighbor (kNN) algorithm for inference. Although this is a common technique in transfer learning, we apply it to the same domain for which the network was trained. Previous works show that neural codes (neuron activations of the last hidden layers) can benefit from the inclusion of classifiers such as support vector machines or random forests. In this work, our proposed hybrid CNN + kNN architecture is evaluated using several image datasets, network topologies and label noise levels. The results show significant accuracy improvements in the inference stage with respect to the standard CNN with noisy labels, especially with relatively large datasets such as CIFAR100. We also verify that applying the ℓ and 2
norm on neural codes is statistically beneficial for this approach.},
keywords = {HispaMus},
pubstate = {published},
tppubtype = {article}
}
We present a hybrid approach to improve the accuracy of Convolutional Neural Networks (CNN) without retraining the model. The proposed architecture replaces the softmax layer by a k-Nearest Neighbor (kNN) algorithm for inference. Although this is a common technique in transfer learning, we apply it to the same domain for which the network was trained. Previous works show that neural codes (neuron activations of the last hidden layers) can benefit from the inclusion of classifiers such as support vector machines or random forests. In this work, our proposed hybrid CNN + kNN architecture is evaluated using several image datasets, network topologies and label noise levels. The results show significant accuracy improvements in the inference stage with respect to the standard CNN with noisy labels, especially with relatively large datasets such as CIFAR100. We also verify that applying the ℓ and 2
norm on neural codes is statistically beneficial for this approach.
norm on neural codes is statistically beneficial for this approach.
Sober-Mira, J.; Calvo-Zaragoza, J.; Rizo, D.; Iñesta, J. M.
Pen-Based Music Document Transcription with Convolutional Neural Networks Book Chapter
In: Fornés, A.; Lamiroy, B. (Ed.): Graphics Recognition. Current Trends and Evolutions, Chapter 6, pp. 71–80, Springer, 2018, ISBN: 978-3-030-02284-6.
Abstract | BibTeX | Tags: HispaMus
@inbook{k400,
title = {Pen-Based Music Document Transcription with Convolutional Neural Networks},
author = {J. Sober-Mira and J. Calvo-Zaragoza and D. Rizo and J. M. Iñesta},
editor = {A. Fornés and B. Lamiroy},
isbn = {978-3-030-02284-6},
year = {2018},
date = {2018-01-01},
booktitle = {Graphics Recognition. Current Trends and Evolutions},
pages = {71--80},
publisher = {Springer},
chapter = {6},
abstract = {The transcription of music sources requires new ways of interacting with musical documents. Assuming that au- tomatic technologies will never guarantee a perfect transcription, our intention is to develop an interactive system in which user and software collaborate to complete the task. Since the use of traditional software for score edition might be tedious, our work studies the interaction by means of electronic pen (e-pen). In our framework, users trace symbols using an e-pen over a digital surface, which provides both the underlying image (offline data) and the drawing made (online data). Using both sources, the system is capable of reaching an error below 4% when recognizing the symbols with a Convolutional Neural Network.},
keywords = {HispaMus},
pubstate = {published},
tppubtype = {inbook}
}
The transcription of music sources requires new ways of interacting with musical documents. Assuming that au- tomatic technologies will never guarantee a perfect transcription, our intention is to develop an interactive system in which user and software collaborate to complete the task. Since the use of traditional software for score edition might be tedious, our work studies the interaction by means of electronic pen (e-pen). In our framework, users trace symbols using an e-pen over a digital surface, which provides both the underlying image (offline data) and the drawing made (online data). Using both sources, the system is capable of reaching an error below 4% when recognizing the symbols with a Convolutional Neural Network.
Calvo-Zaragoza, J.; Castellanos, F. J.; Vigliensoni, G.; Fujinaga, I.
Deep Neural Networks for Document Processing of Music Score Images Journal Article
In: Applied Sciences, vol. 8, no. 5, pp. 654, 2018.
@article{k427,
title = {Deep Neural Networks for Document Processing of Music Score Images},
author = {J. Calvo-Zaragoza and F. J. Castellanos and G. Vigliensoni and I. Fujinaga},
year = {2018},
date = {2018-01-01},
urldate = {2018-01-01},
journal = {Applied Sciences},
volume = {8},
number = {5},
pages = {654},
keywords = {HispaMus},
pubstate = {published},
tppubtype = {article}
}
Gallego, A. J.; Calvo-Zaragoza, J.; Valero-Mas, J. J.; Rico-Juan, J. R.
Clustering-based k-nearest neighbor classification for large-scale data with neural codes representation Journal Article
In: Pattern Recognition, vol. 74, pp. 531-543, 2018.
@article{k378,
title = {Clustering-based k-nearest neighbor classification for large-scale data with neural codes representation},
author = {A. J. Gallego and J. Calvo-Zaragoza and J. J. Valero-Mas and J. R. Rico-Juan},
year = {2018},
date = {2018-01-01},
urldate = {2018-01-01},
journal = {Pattern Recognition},
volume = {74},
pages = {531-543},
keywords = {TIMuL},
pubstate = {published},
tppubtype = {article}
}
2017
Sober-Mira, J.; Calvo-Zaragoza, J.; Rizo, D.; Iñesta, J. M.
Pen-based music document transcription Proceedings Article
In: Proceedings of GREC 2017, pp. 21–22, IEEE computer society, Kyoto (Japan), 2017, ISBN: 978-1-5386-3586-5.
@inproceedings{k381,
title = {Pen-based music document transcription},
author = {J. Sober-Mira and J. Calvo-Zaragoza and D. Rizo and J. M. Iñesta},
url = {https://grfia.dlsi.ua.es/repositori/grfia/pubs/381/3586c021.pdf},
isbn = {978-1-5386-3586-5},
year = {2017},
date = {2017-11-01},
booktitle = {Proceedings of GREC 2017},
pages = {21--22},
publisher = {IEEE computer society},
address = {Kyoto (Japan)},
keywords = {TIMuL},
pubstate = {published},
tppubtype = {inproceedings}
}
Calvo-Zaragoza, J.; Gallego, A. J.; Pertusa, A.
Recognition of Handwritten Music Symbols with Convolutional Neural Codes Proceedings Article
In: 14th IAPR International Conference on Document Analysis and Recognition (ICDAR), pp. 691–696, Kyoto, Japan, 2017.
BibTeX | Tags: GRE16-14, TIMuL
@inproceedings{k376,
title = {Recognition of Handwritten Music Symbols with Convolutional Neural Codes},
author = {J. Calvo-Zaragoza and A. J. Gallego and A. Pertusa},
year = {2017},
date = {2017-11-01},
urldate = {2017-11-01},
booktitle = {14th IAPR International Conference on Document Analysis and Recognition (ICDAR)},
pages = {691--696},
address = {Kyoto, Japan},
keywords = {GRE16-14, TIMuL},
pubstate = {published},
tppubtype = {inproceedings}
}
Calvo-Zaragoza, J.; Valero-Mas, J. J.; Pertusa, A
End-To-End Optical Music Recognition using Neural Networks Proceedings Article
In: Proc. of International Society for Music Information Retrieval Conference (ISMIR), Suzhou, China, 2017.
Abstract | BibTeX | Tags: GRE16-14, TIMuL
@inproceedings{k374,
title = {End-To-End Optical Music Recognition using Neural Networks},
author = {J. Calvo-Zaragoza and J. J. Valero-Mas and A Pertusa},
year = {2017},
date = {2017-10-01},
booktitle = {Proc. of International Society for Music Information Retrieval Conference (ISMIR)},
address = {Suzhou, China},
abstract = {This work addresses the Optical Music Recognition (OMR) task in an end-to-end fashion using neural net- works. The proposed architecture is based on a Recurrent Convolutional Neural Network topology that takes as input an image of a monophonic score and retrieves a sequence of music symbols as output. In the first stage, a series of convolutional filters are trained to extract meaningful fea- tures of the input image, and then a recurrent block models the sequential nature of music. The system is trained us- ing a Connectionist Temporal Classification loss function, which avoids the need for a frame-by-frame alignment be- tween the image and the ground-truth music symbols. Ex- perimentation has been carried on a set of 90,000 synthetic monophonic music scores with more than 50 different pos- sible labels. Results obtained depict classification error rates around 2 % at symbol level, thus proving the po- tential of the proposed end-to-end architecture for OMR. The source code, dataset, and trained models are publicly released for reproducible research and future comparison purposes.},
keywords = {GRE16-14, TIMuL},
pubstate = {published},
tppubtype = {inproceedings}
}
This work addresses the Optical Music Recognition (OMR) task in an end-to-end fashion using neural net- works. The proposed architecture is based on a Recurrent Convolutional Neural Network topology that takes as input an image of a monophonic score and retrieves a sequence of music symbols as output. In the first stage, a series of convolutional filters are trained to extract meaningful fea- tures of the input image, and then a recurrent block models the sequential nature of music. The system is trained us- ing a Connectionist Temporal Classification loss function, which avoids the need for a frame-by-frame alignment be- tween the image and the ground-truth music symbols. Ex- perimentation has been carried on a set of 90,000 synthetic monophonic music scores with more than 50 different pos- sible labels. Results obtained depict classification error rates around 2 % at symbol level, thus proving the po- tential of the proposed end-to-end architecture for OMR. The source code, dataset, and trained models are publicly released for reproducible research and future comparison purposes.
Sober-Mira, J.; Calvo-Zaragoza, J.; Rizo, D.; Iñesta, J. M.
Multimodal Recognition for Music Document Transcription Proceedings Article
In: Proceedings of MML 2017, pp. 67–72, Barcelona, 2017.
@inproceedings{k380,
title = {Multimodal Recognition for Music Document Transcription},
author = {J. Sober-Mira and J. Calvo-Zaragoza and D. Rizo and J. M. Iñesta},
url = {https://grfia.dlsi.ua.es/repositori/grfia/pubs/380/mml17proceedings-67.pdf},
year = {2017},
date = {2017-10-01},
booktitle = {Proceedings of MML 2017},
pages = {67--72},
address = {Barcelona},
keywords = {},
pubstate = {published},
tppubtype = {inproceedings}
}
Valero-Mas, J. J.; Calvo-Zaragoza, J.; Rico-Juan, J. R.; Iñesta, J. M.
A study of Prototype Selection algorithms for Nearest Neighbour in class-imbalanced problems Proceedings Article
In: Alexandre, J. S. Sánchez L. A.; Rodrigues, J. M. F. (Ed.): Proceedings of the 8th Iberian Conference on Pattern Recognition and Image Analysis (IbPRIA), pp. 335–343, Springer, Faro, Portugal, 2017, ISBN: 978-3-319-58837-7.
@inproceedings{k362,
title = {A study of Prototype Selection algorithms for Nearest Neighbour in class-imbalanced problems},
author = {J. J. Valero-Mas and J. Calvo-Zaragoza and J. R. Rico-Juan and J. M. Iñesta},
editor = {J. S. Sánchez L. A. Alexandre and J. M. F. Rodrigues},
url = {https://grfia.dlsi.ua.es/repositori/grfia/pubs/362/CameraReady.pdf},
isbn = {978-3-319-58837-7},
year = {2017},
date = {2017-06-01},
booktitle = {Proceedings of the 8th Iberian Conference on Pattern Recognition and Image Analysis (IbPRIA)},
pages = {335--343},
publisher = {Springer},
address = {Faro, Portugal},
keywords = {TIMuL},
pubstate = {published},
tppubtype = {inproceedings}
}
Rizo, D.; Calvo-Zaragoza, J.; Iñesta, J. M.; Fujinaga, I.
About agnostic representation of musical documents for Optical Music Recognition Proceedings Article
In: Music Encoding Conference, Tours, 2017, 2017.
@inproceedings{k369,
title = {About agnostic representation of musical documents for Optical Music Recognition},
author = {D. Rizo and J. Calvo-Zaragoza and J. M. Iñesta and I. Fujinaga},
year = {2017},
date = {2017-05-01},
booktitle = {Music Encoding Conference, Tours, 2017},
keywords = {TIMuL},
pubstate = {published},
tppubtype = {inproceedings}
}
Calvo-Zaragoza, J.; Oncina, J.
Recognition of Pen-based Music Notation with Finite-State Machines Journal Article
In: Expert Systems With Applications, vol. 72, pp. 395-406, 2017.
@article{k358,
title = {Recognition of Pen-based Music Notation with Finite-State Machines},
author = {J. Calvo-Zaragoza and J. Oncina},
url = {https://grfia.dlsi.ua.es/repositori/grfia/pubs/358/recognition-pen-based.pdf},
year = {2017},
date = {2017-04-01},
journal = {Expert Systems With Applications},
volume = {72},
pages = {395-406},
keywords = {TIMuL},
pubstate = {published},
tppubtype = {article}
}
Calvo-Zaragoza, J.; Oncina, J.
An efficient approach for Interactive Sequential Pattern Recognition Journal Article
In: Pattern Recognition, vol. 64, pp. 295-304, 2017.
@article{k359,
title = {An efficient approach for Interactive Sequential Pattern Recognition},
author = {J. Calvo-Zaragoza and J. Oncina},
url = {https://grfia.dlsi.ua.es/repositori/grfia/pubs/359/efficient-approach-sequential.pdf},
year = {2017},
date = {2017-04-01},
journal = {Pattern Recognition},
volume = {64},
pages = {295-304},
keywords = {TIMuL},
pubstate = {published},
tppubtype = {article}
}
Valero-Mas, J. J.; Calvo-Zaragoza, J.; Rico-Juan, J. R.; Iñesta, J. M.
An Experimental Study on Rank Methods for Prototype Selection Journal Article
In: Soft Computing, vol. 21, no. 19, pp. 5703-–5715, 2017, ISSN: 1432-7643.
Abstract | Links | BibTeX | Tags: TIMuL
@article{k339,
title = {An Experimental Study on Rank Methods for Prototype Selection},
author = {J. J. Valero-Mas and J. Calvo-Zaragoza and J. R. Rico-Juan and J. M. Iñesta},
url = {https://grfia.dlsi.ua.es/repositori/grfia/pubs/339/SOCO-RankMethods-2016.pdf},
issn = {1432-7643},
year = {2017},
date = {2017-01-01},
journal = {Soft Computing},
volume = {21},
number = {19},
pages = {5703-–5715},
abstract = {Prototype selection is one of the most popular approaches for addressing the low efficiency issue typically found in the well-known k-Nearest Neighbour classification rule. These techniques select a representative subset from an original collection of prototypes with the premise of main- taining the same classification accuracy. Most recently, rank methods have been proposed as an alternative to develop new selection strategies. Following a certain heuristic, these methods sort the elements of the initial collection accord- ing to their relevance and then select the best possible subset by means of a parameter representing the amount of data to maintain. Due to the relative novelty of these methods, their performance and competitiveness against other strategies is still unclear. This work performs an exhaustive experimental study of such methods for prototype selection. A represen- tative collection of both classic and sophisticated algorithms are compared to the aforementioned techniques in a num- ber of datasets, including different levels of induced noise. Results report the remarkable competitiveness of these rank methods as well as their excellent trade-off between proto- type reduction and achieved accuracy.},
keywords = {TIMuL},
pubstate = {published},
tppubtype = {article}
}
Prototype selection is one of the most popular approaches for addressing the low efficiency issue typically found in the well-known k-Nearest Neighbour classification rule. These techniques select a representative subset from an original collection of prototypes with the premise of main- taining the same classification accuracy. Most recently, rank methods have been proposed as an alternative to develop new selection strategies. Following a certain heuristic, these methods sort the elements of the initial collection accord- ing to their relevance and then select the best possible subset by means of a parameter representing the amount of data to maintain. Due to the relative novelty of these methods, their performance and competitiveness against other strategies is still unclear. This work performs an exhaustive experimental study of such methods for prototype selection. A represen- tative collection of both classic and sophisticated algorithms are compared to the aforementioned techniques in a num- ber of datasets, including different levels of induced noise. Results report the remarkable competitiveness of these rank methods as well as their excellent trade-off between proto- type reduction and achieved accuracy.
Calvo-Zaragoza, J.; Vigliensoni, G.; Fujinaga, I.
A machine learning framework for the categorization of elements in images of musical documents Proceedings Article
In: Proceedings of the Third International Conference on Technologies for Music Notation and Representation, 2017.
@inproceedings{k360,
title = {A machine learning framework for the categorization of elements in images of musical documents},
author = {J. Calvo-Zaragoza and G. Vigliensoni and I. Fujinaga},
url = {https://grfia.dlsi.ua.es/repositori/grfia/pubs/360/tenor-unified-categorization.pdf},
year = {2017},
date = {2017-01-01},
booktitle = {Proceedings of the Third International Conference on Technologies for Music Notation and Representation},
keywords = {TIMuL},
pubstate = {published},
tppubtype = {inproceedings}
}
Calvo-Zaragoza, J.; Vigliensoni, G.; Fujinaga, I.
Pixel-wise Binarization of Musical Documents with Convolutional Neural Networks Proceedings Article
In: Proceedings of the 15th IAPR International Conference on Machine Vision Applications, 2017.
@inproceedings{k361,
title = {Pixel-wise Binarization of Musical Documents with Convolutional Neural Networks},
author = {J. Calvo-Zaragoza and G. Vigliensoni and I. Fujinaga},
url = {https://grfia.dlsi.ua.es/repositori/grfia/pubs/361/pixel-wise-binarization.pdf},
year = {2017},
date = {2017-01-01},
booktitle = {Proceedings of the 15th IAPR International Conference on Machine Vision Applications},
keywords = {TIMuL},
pubstate = {published},
tppubtype = {inproceedings}
}
Calvo-Zaragoza, J.; Pertusa, A.; Oncina, J.
Staff-line detection and removal using a convolutional neural network Journal Article
In: Machine Vision and Applications, pp. 1-10, 2017, ISSN: 1432-1769.
Abstract | BibTeX | Tags: TIMuL
@article{k365,
title = {Staff-line detection and removal using a convolutional neural network},
author = {J. Calvo-Zaragoza and A. Pertusa and J. Oncina},
issn = {1432-1769},
year = {2017},
date = {2017-01-01},
urldate = {2017-01-01},
journal = {Machine Vision and Applications},
pages = {1-10},
abstract = {Staff-line removal is an important preprocessing stage for most optical music recognition systems. Common procedures to solve this task involve image processing techniques. In contrast to these traditional methods based on hand-engineered transformations, the problem can also be approached as a classification task in which each pixel is labeled as either staff or symbol, so that only those that belong to symbols are kept in the image. In order to perform this classification, we propose the use of convolutional neural networks, which have demonstrated an outstanding performance in image retrieval tasks. The initial features of each pixel consist of a square patch from the input image centered at that pixel. The proposed network is trained by using a dataset which contains pairs of scores with and without the staff lines. Our results in both binary and grayscale images show that the proposed technique is very accurate, outperforming both other classifiers and the state-of-the-art strategies considered. In addition, several advantages of the presented methodology with respect to traditional procedures proposed so far are discussed.},
keywords = {TIMuL},
pubstate = {published},
tppubtype = {article}
}
Staff-line removal is an important preprocessing stage for most optical music recognition systems. Common procedures to solve this task involve image processing techniques. In contrast to these traditional methods based on hand-engineered transformations, the problem can also be approached as a classification task in which each pixel is labeled as either staff or symbol, so that only those that belong to symbols are kept in the image. In order to perform this classification, we propose the use of convolutional neural networks, which have demonstrated an outstanding performance in image retrieval tasks. The initial features of each pixel consist of a square patch from the input image centered at that pixel. The proposed network is trained by using a dataset which contains pairs of scores with and without the staff lines. Our results in both binary and grayscale images show that the proposed technique is very accurate, outperforming both other classifiers and the state-of-the-art strategies considered. In addition, several advantages of the presented methodology with respect to traditional procedures proposed so far are discussed.
Gallego, A. J.; Calvo-Zaragoza, J.
Staff-line removal with Selectional Auto-Encoders Journal Article
In: Expert Systems With Applications, vol. 89, pp. 138 - 148, 2017, ISSN: 0957-4174.
Abstract | Links | BibTeX | Tags: TIMuL
@article{k372,
title = {Staff-line removal with Selectional Auto-Encoders},
author = {A. J. Gallego and J. Calvo-Zaragoza},
url = {https://grfia.dlsi.ua.es/repositori/grfia/pubs/372/staff-line-removal.pdf},
issn = {0957-4174},
year = {2017},
date = {2017-01-01},
urldate = {2017-01-01},
journal = {Expert Systems With Applications},
volume = {89},
pages = {138 - 148},
abstract = {Staff-line removal is an important preprocessing stage as regards most Optical Music Recognition systems. The common procedures employed to carry out this task involve image processing techniques. In contrast to these traditional methods, which are based on hand-engineered transformations, the problem can also be approached from a machine learning point of view if representative examples of the task are provided. We propose doing this through the use of a new approach involving auto-encoders, which select the appropriate features of an input feature set (Selectional Auto-Encoders). Within the context of the problem at hand, the model is trained to select those pixels of a given image that belong to a musical symbol, thus removing the lines of the staves. Our results show that the proposed technique is quite competitive and significantly outperforms the other state-of-art strategies considered, particularly when dealing with grayscale input images.},
keywords = {TIMuL},
pubstate = {published},
tppubtype = {article}
}
Staff-line removal is an important preprocessing stage as regards most Optical Music Recognition systems. The common procedures employed to carry out this task involve image processing techniques. In contrast to these traditional methods, which are based on hand-engineered transformations, the problem can also be approached from a machine learning point of view if representative examples of the task are provided. We propose doing this through the use of a new approach involving auto-encoders, which select the appropriate features of an input feature set (Selectional Auto-Encoders). Within the context of the problem at hand, the model is trained to select those pixels of a given image that belong to a musical symbol, thus removing the lines of the staves. Our results show that the proposed technique is quite competitive and significantly outperforms the other state-of-art strategies considered, particularly when dealing with grayscale input images.
2016
Rizo, D.; Calvo-Zaragoza, J.; Iñesta, J. M.; Illescas, P. R.
Hidden Markov Models for Functional Analysis Proceedings Article
In: Music and Machine Learning Workshop, Riva del Garda, 2016.
@inproceedings{k370,
title = {Hidden Markov Models for Functional Analysis},
author = {D. Rizo and J. Calvo-Zaragoza and J. M. Iñesta and P. R. Illescas},
year = {2016},
date = {2016-09-01},
booktitle = {Music and Machine Learning Workshop, Riva del Garda},
keywords = {TIMuL},
pubstate = {published},
tppubtype = {inproceedings}
}
Valero-Mas, J. J.; Calvo-Zaragoza, J.; Rico-Juan, J. R.
On the suitability of Prototype Selection methods for kNN classification with distributed data Journal Article
In: Neurocomputing, vol. 203, pp. 150-160, 2016.
@article{k341,
title = {On the suitability of Prototype Selection methods for kNN classification with distributed data},
author = {J. J. Valero-Mas and J. Calvo-Zaragoza and J. R. Rico-Juan},
url = {https://grfia.dlsi.ua.es/repositori/grfia/pubs/341/SuitabilityPSDistributedScenarios.pdf},
year = {2016},
date = {2016-08-01},
journal = {Neurocomputing},
volume = {203},
pages = {150-160},
keywords = {TIMuL},
pubstate = {published},
tppubtype = {article}
}
Calvo-Zaragoza, J.; Rizo, D.; Iñesta, J. M.
Two (note) heads are better than one: pen-based multimodal interaction with music scores Proceedings Article
In: Devaney, J. (Ed.): 17th International Society for Music Information Retrieval Conference, pp. 509-514, New York City, 2016, ISBN: 978-0-692-75506-8.
@inproceedings{k345,
title = {Two (note) heads are better than one: pen-based multimodal interaction with music scores},
author = {J. Calvo-Zaragoza and D. Rizo and J. M. Iñesta},
editor = {J. Devaney},
url = {https://grfia.dlsi.ua.es/repositori/grfia/pubs/345/two-note-heads.pdf},
isbn = {978-0-692-75506-8},
year = {2016},
date = {2016-08-01},
booktitle = {17th International Society for Music Information Retrieval Conference},
pages = {509-514},
address = {New York City},
keywords = {TIMuL},
pubstate = {published},
tppubtype = {inproceedings}
}
Calvo-Zaragoza, J.; Oncina, J.; Higuera, C. De La
Computing the Expected Edit Distance from a String to a PFA Proceedings Article
In: Han, Yo-Sub; Salomaa, Kai (Ed.): 21st International Conference Implementation and Application of Automata, pp. 39-50, Springer, 2016.
@inproceedings{k342,
title = {Computing the Expected Edit Distance from a String to a PFA},
author = {J. Calvo-Zaragoza and J. Oncina and C. De La Higuera},
editor = {Yo-Sub Han and Kai Salomaa},
url = {https://grfia.dlsi.ua.es/repositori/grfia/pubs/342/distance-string-pfa.pdf},
year = {2016},
date = {2016-07-01},
urldate = {2016-07-01},
booktitle = {21st International Conference Implementation and Application of Automata},
pages = {39-50},
publisher = {Springer},
keywords = {TIMuL},
pubstate = {published},
tppubtype = {inproceedings}
}
Calvo-Zaragoza, J.; Micó, L.; Oncina, J.
Music staff removal with supervised pixel classification Journal Article
In: International Journal on Document Analysis and Recognition, vol. 19, no. 3, pp. 211-219, 2016, ISSN: 1433-2833.
@article{k336,
title = {Music staff removal with supervised pixel classification},
author = {J. Calvo-Zaragoza and L. Micó and J. Oncina},
url = {https://grfia.dlsi.ua.es/repositori/grfia/pubs/336/classification-approach-staff.pdf},
issn = {1433-2833},
year = {2016},
date = {2016-01-01},
journal = {International Journal on Document Analysis and Recognition},
volume = {19},
number = {3},
pages = {211-219},
keywords = {TIMuL},
pubstate = {published},
tppubtype = {article}
}
Calvo-Zaragoza, J.; Valero-Mas, J. J.; Rico-Juan, J. R.
Prototype Generation on Structural Data using Dissimilarity Space Representation Journal Article
In: Neural Computing and Applications, 2016.
@article{k337,
title = {Prototype Generation on Structural Data using Dissimilarity Space Representation},
author = {J. Calvo-Zaragoza and J. J. Valero-Mas and J. R. Rico-Juan},
url = {https://grfia.dlsi.ua.es/repositori/grfia/pubs/337/prototype-generation-structural.pdf},
year = {2016},
date = {2016-01-01},
journal = {Neural Computing and Applications},
keywords = {TIMuL},
pubstate = {published},
tppubtype = {article}
}
Calvo-Zaragoza, J.; Valero-Mas, J. J.; Rico-Juan, J. R.
Selecting promising classes from generated data for an efficient multi-class NN classification Journal Article
In: Soft Computing, 2016.
@article{k340,
title = {Selecting promising classes from generated data for an efficient multi-class NN classification},
author = {J. Calvo-Zaragoza and J. J. Valero-Mas and J. R. Rico-Juan},
url = {https://grfia.dlsi.ua.es/repositori/grfia/pubs/340/selecting-promising-classes.pdf},
year = {2016},
date = {2016-01-01},
journal = {Soft Computing},
keywords = {TIMuL},
pubstate = {published},
tppubtype = {article}
}
Calvo-Zaragoza, J.; Toselli, A. H.; Vidal, E.
Early Handwritten Music Recognition with Hidden Markov Models Proceedings Article
In: 15th International Conference on Frontiers in Handwriting Recognition, 2016.
@inproceedings{k350,
title = {Early Handwritten Music Recognition with Hidden Markov Models},
author = {J. Calvo-Zaragoza and A. H. Toselli and E. Vidal},
url = {https://grfia.dlsi.ua.es/repositori/grfia/pubs/350/musicNoteRecogIcfhr16.pdf},
year = {2016},
date = {2016-01-01},
booktitle = {15th International Conference on Frontiers in Handwriting Recognition},
keywords = {TIMuL},
pubstate = {published},
tppubtype = {inproceedings}
}