Publications
Publications are generated from a single BibTeX source. Use the controls to filter by type or change the sort order. Everything stays visible with JavaScript disabled.
-
Karmim, Y., Pino, R., Contreras, H., Lira, H., Cifuentes, S., Escoffier, S., Martí, L., Seddah, D., & Barrière, V. (2026). Leveraging Wikidata for Geographically Informed Sociocultural Bias Dataset Creation: Application to Latin America. In P. Chen, V. Zouhar, H. Hu, S. Khanuja, W. Zhu, B. Haddow, A. Birch, A. F. Aji, R. Sennrich, & S. Hooker (Eds.), Proceedings of the Workshop on Multilingual Multicultural Evaluation of the 19th Conference of the European Chapter of the Association for Computational Linguistics (EACL’2026). https://inria.hal.science/hal-05510068
@inproceedings{karmimleveraging2026, title = {Leveraging Wikidata for Geographically Informed Sociocultural Bias Dataset Creation: {A}pplication to {L}atin {A}merica}, author = {Karmim, Yannis and Pino, Renato and Contreras, Hernan and Lira, Hernan and Cifuentes, Sebastien and Escoffier, Simon and Mart\'{i}, Luis and Seddah, Djam{\'e} and Barri{\`e}re, Valentin}, year = {2026}, month = mar, booktitle = {Proceedings of the Workshop on Multilingual Multicultural Evaluation of the 19th Conference of the European Chapter of the Association for Computational Linguistics (EACL'2026)}, address = {Rabbat, Morocco}, url = {https://inria.hal.science/hal-05510068}, editor = {Chen, Pinzhen and Zouhar, Vil\'{e}m and Hu, Hanxu and Khanuja, Simran and Zhu, Wenhao and Haddow, Barry and Birch, Alexandra and Aji, Alham Fikri and Sennrich, Rico and Hooker, Sara}, hal_id = {hal-05510068}, hal_version = {v1}, eprint = {hal-05510068}, eprinttype = {hal} } -
Vinayagame, V., Senay, G., & Martí, L. (2025). MATATA: Weakly Supervised End-to-End MAthematical Tool-Augmented Reasoning for Tabular Applications. Document Analysis and Recognition – ICDAR 2025: 19th International Conference, Wuhan, China, September 16–21, 2025, Proceedings, Part I, 428–467. https://doi.org/10.1007/978-3-032-04614-7_24
Business documents often contain substantial tabular and textual information with numerical values, requiring mathematical reasoning for effective document understanding. While small language models (SLMs) still struggle at this task, tool-augmented multi-step agents perform better, at the cost of relying on closed-source or larger models, external data, or extensive prompt engineering. This work introduces MATATA, a novel weakly supervised end-to-end approach to training multi-step reasoning language agents for document tabular applications. MATATA presents an annotation-free paradigm for each agent to enhance 3.8B/8B SLMs. During its two-stage training, MATATA uses the outcome of the multi-step reasoning chain as weak supervision. This approach avoids having to individually supervise each intermediate agent in the reasoning chain. By employing an adaptive planner and shared tools across different datasets, MATATA shows robust performance. Experiments demonstrate that MATATA achieves state-of-the-art on FinQA, and on TAT-QA among reasoning methods based on open-source SLMs. Although being SLM-based, MATATA closely matches GPT-4-based frameworks on TabMWP. This novel weakly supervised approach enables training an end-to-end multi-step reasoning agent without intermediate supervision, supporting future developments of cost-effective powerful agentic systems.
@inproceedings{10.1007/978-3-032-04614-7_24, author = {Vinayagame, Vishnou and Senay, Gregory and Mart\'{i}, Luis}, title = {MATATA: {W}eakly Supervised End-to-End MAthematical Tool-Augmented Reasoning for Tabular Applications}, year = {2025}, isbn = {978-3-032-04613-0}, publisher = {Springer-Verlag}, address = {Berlin, Heidelberg}, url = {https://doi.org/10.1007/978-3-032-04614-7_24}, doi = {10.1007/978-3-032-04614-7_24}, booktitle = {Document Analysis and Recognition – ICDAR 2025: 19th International Conference, Wuhan, China, September 16–21, 2025, Proceedings, Part I}, pages = {428--467}, keywords = {Document Understanding, Tabular Mathematical Reasoning, Small Language Models, Weak Supervision}, location = {Wuhan, China} } -
Vinayagame, V., Senay, G., & Martí, L. (2024). Tool-Augmented Compositional Reasoning LLMs with Weak Supervision: A Scalable Approach to Reduce Human Efforts in Agent Customization. BayLearn 2024: Bay Area Machine Learning Symposimum.
Mathematical reasoning capabilities are increasing with tool-augmented language agents, but methods often rely on proprietary models to generate trajectories for training or human efforts for prompt engineering. This work introduces a progressive refinement learning paradigm through self-annotation and weak-supervision. By updating the model’s beliefs and evolving from human inputs, the reliance on human supervision and stronger teacher models is minimized, jointly reducing the need to adapt prompting to the models.
@inproceedings{vishnou-2024, author = {Vinayagame, Vishnou and Senay, Gregory and Mart\'{i}, Luis}, booktitle = {BayLearn 2024: Bay Area Machine Learning Symposimum}, title = {Tool-Augmented Compositional Reasoning LLMs with Weak Supervision: A Scalable Approach to Reduce Human Efforts in Agent Customization}, year = {2024}, location = {San Francisco, CA, USA} } -
Begun, A. P., DeRose, S., Jaffri, T., Martí, L., Palmer, M. B., Paoli, J., Pavlopoulou, C., Pricoiu, E., Sarangi, S., Sawicki, M., Shehadeh, M., Taron, M., Toprani, B., Wadia, Z. R., Watson, D., White, E., Fan, J. Y., Gupta, K., Hoang, A. M., … Zhou, X. (2024). Cross-Document Intelligent Authoring and Processing, with Arbitration for Semantically-Annotated Documents (Patentus) [Patentus]. https://patents.google.com/patent/US20220245335A1/
Machine learning, artificial intelligence, and other computer-implemented methods are used to identify various semantically important chunks in documents, automatically label them with appropriate datatypes and semantic roles, and use this enhanced information to assist authors and to support downstream processes. Chunk locations, datatypes, and semantic roles can often be automatically determined from what is here called “context”, to wit, the combination of their formatting, structure, and content; those of adjacent or nearby content; overall patterns of occurrence in a document, and similarities of all these things across documents (mainly but not exclusively among documents in the same document set). Similarity is not limited to exact or fuzzy string or property comparisons, but may include similarity of natural language grammatical structure, ML (machine learning) techniques such as measuring similarity of word, chunk, and other embeddings, and the datatypes and semantic roles of previously-identified chunks.
@patent{Begun2022-bx, author = {Begun, Andrew Paul and DeRose, Steven and Jaffri, Taqi and Mart\'{i}, Luis and Palmer, Michael B and Paoli, Jean and Pavlopoulou, Christina and Pricoiu, Elena and Sarangi, Swagatika and Sawicki, Martin and Shehadeh, Manar and Taron, Michael and Toprani, Bhaven and Wadia, Zubin Rustom and Watson, David and White, Eric and Fan, Joshua Yongshin and Gupta, Kush and Hoang, Andrew Minh and Liu, Zhanlin and Paliakkara, Jerome George and Wu, Zhaofeng and Zhang, Yue and Zhou, Xiaoquan}, holder = {Docugami~Inc.}, location = {USA}, month = mar, number = {20220245335:A1}, title = {Cross-Document Intelligent Authoring and Processing, with Arbitration for Semantically-Annotated Documents}, type = {patentus}, url = {https://patents.google.com/patent/US20220245335A1/}, year = {2024}, bdsk-url-1 = {https://patents.google.com/patent/US20220245335A1/} } -
Salgado Pereira, C., Mota Dias, D., Martí, L., & Vellasco, M. M. B. R. (2023). A Multi-Objective Decomposition Optimization Method for Refinery Crude Oil Scheduling through Genetic Programming. Proceedings of the Companion Conference on Genetic and Evolutionary Computation, 1972–1980. https://doi.org/10.1145/3583133.3596313
This paper proposes an evolutionary algorithm integrating genetic programming and a decomposition-based multi-objective algorithm to address a crude oil refinery scheduling problem. Four objectives are modelled, two related to maintaining the crude oil processing level, and the other two aim to keep the refinery operations as smooth as possible. The proposed method, Constrained-Decomposition of Quantum-Inspired Grammar-based Linear Genetic Programming (C-DQIGLGP), uses Quantum-Inspired Grammar-based Linear Genetic Programming (QIGLGP), replacing its hierarchical approach for the objectives with a multi-objective decomposition-based one. To achieve this goal, QIGLGP was profoundly modified regarding sorting individuals, updating the population, and applying the evolutionary operator. Individuals whose objective values related to processing level are under a predefined limit are better ranked. We compare the results of C-DQIGLGP for five scenarios from a real refinery to those obtained by QIGLGP and Constrained Non-dominated Sort QIGLGP (C-NSQIGLGP), from literature, demonstrating the better performance of C-DQIGLGP for all cases.
@inproceedings{10.1145/3583133.3596313, address = {New York, NY, USA}, author = {Salgado Pereira, Cristiane and Mota Dias, Douglas and Mart\'{i}, Luis and Vellasco, Marley~M.~B.~R.}, booktitle = {Proceedings of the Companion Conference on Genetic and Evolutionary Computation}, doi = {10.1145/3583133.3596313}, isbn = {979-8-4007-0120-7}, keywords = {genetic programming, decomposition, quantum-inspired algorithm, refinery scheduling, evolutionary multi-objective optimization}, location = {Lisbon, Portugal}, numpages = {9}, pages = {1972--1980}, publisher = {Association for Computing Machinery}, series = {GECCO'23 Companion}, title = {A Multi-Objective Decomposition Optimization Method for Refinery Crude Oil Scheduling through {G}enetic {P}rogramming}, url = {https://doi.org/10.1145/3583133.3596313}, year = {2023}, bdsk-url-1 = {https://doi.org/10.1145/3583133.3596313} } -
de Moraes, R. F., Evangelista, R. dos S., Pereira, A. L. da S., Toledo, Y. P., Fernandes, L. A. F., & Martí, L. (2023). Heuristics to reduce linear combinations of activation functions to improve image classification. 2023 36th SIBGRAPI Conference on Graphics, Patterns and Images (SIBGRAPI), 169–174. https://doi.org/10.1109/SIBGRAPI59091.2023.10347043
Image classification is one of the classical problems in computer vision, and CNNs (Convolutional Neural Networks) are widely used for this task. However, the choice of a CNN can vary depending on the chosen dataset. In this context, we have trainable activation functions that are crucial in CNNs and adapt to the data. One technique for constructing these functions is to write them as a linear combination of other activation functions, where the coefficients of this combination are learned during training. However, if we have a large number of activation functions to combine, the computational cost can be very high, and manually testing and choosing these functions may be impractical, depending on the number of available activation functions. To alleviate the difficulty of choosing which activation functions should be part of the linear combination, we propose two heuristics: Linear Combination Approximator by Coefficients (LCAC) and Major and Uniform Coefficient Extractor (MUCE). Our heuristics provide an efficient selection of a subset of activation functions so that their results are better or equivalent to the linear combination that uses all 34 available activation functions in our experiments (C34), considering the image classification problem. Compared to the C34 function, the LCAC function was better or equivalent in 62.5%, and the MUCE function in 87.5% of the conducted experiments.
@inproceedings{10347043, author = {de Moraes, Rog\'{e}rio Ferreira and Evangelista, Raphael dos S. and Pereira, Andre Luiz da S. and Toledo, Yanexis Pupo and Fernandes, Leandro A. F. and Mart\'{i}, Luis}, booktitle = {2023 36th SIBGRAPI Conference on Graphics, Patterns and Images (SIBGRAPI)}, doi = {10.1109/SIBGRAPI59091.2023.10347043}, hal_id = {hal-04396390}, hal_version = {v1}, issn = {2377-5416}, keywords = {Training ; Graphics ; Computer vision ; Computational efficiency ; Convolutional neural networks ; Task analysis ; Image classification}, location = {Rio Grande, Brazil}, month = nov, pages = {169--174}, publisher = {IEEE}, title = {Heuristics to reduce linear combinations of activation functions to improve image classification}, year = {2023}, bdsk-url-1 = {https://doi.org/10.1109/SIBGRAPI59091.2023.10347043} } -
Marinho, W., Clua, E. W., Martí, L., & Marinho, K. (2023). Predicting Item Response Theory Parameters Using Question Statements Texts. LAK23: 13th International Learning Analytics and Knowledge Conference, 1–10. https://doi.org/10.1145/3576050.3576139
Recently, new Neural Language Models pre-trained on a massive corpus of texts are available. These models encode statistical features of the languages through their parameters, creating better word vector representations that allow the training of neural networks with smaller sample sets. In this context, we investigate the application of these models to predict Item Response Theory parameters in multiple choice questions. More specifically, we apply our models for the Brazilian National High School Exam (ENEM) questions using the text of their statements and propose a novel optimization target for regression: Item Characteristic Curve. The architecture employed could predict the difficulty parameter b of the ENEM 2020 and 2021 items with a mean absolute error of 70 points. Calculating the IRT score in each knowledge area of the exam for a sample of 100,000 students, we obtained a mean absolute below 40 points for all knowledge areas. Considering only the top quartile, the exam’s main target of interest, the average error was less than 30 points for all areas, being the majority lower than 15 points. Such performance allows predicting parameters on newly created questions, composing mock tests for student training, and analyzing their performance with excellent precision, dispensing with the need for costly item calibration pre-test step.
@inproceedings{10.1145/3576050.3576139, address = {New York, NY, USA}, author = {Marinho, Wemerson and Clua, Esteban Walter and Mart\'{i}, Luis and Marinho, Karla}, booktitle = {LAK23: 13th International Learning Analytics and Knowledge Conference}, doi = {10.1145/3576050.3576139}, isbn = {9781450398657}, keywords = {ENEM, Item response theory, Text regression, Question difficulty prediction}, location = {Arlington, TX, USA}, numpages = {10}, pages = {1--10}, publisher = {Association for Computing Machinery}, series = {LAK2023}, title = {Predicting Item Response Theory Parameters Using Question Statements Texts}, url = {https://doi.org/10.1145/3576050.3576139}, year = {2023}, bdsk-url-1 = {https://doi.org/10.1145/3576050.3576139} } -
Begun, A., DeRose, S., Jaffri, T., Martí, L., Palmer, M., Paoli, J., Pavlopoulou, C., Pricoiu, E., Sarangi, S., Sawicki, M., Shehadeh, M., Taron, M., Toprani, B., Wadia, Z. R., Watson, D., White, E., Fan, J. Y., Gupta, K., Hoang, A. M., … Zhou, X. (2023). Automatically identifying chunks in sets of documents (Patent) [Patent]. https://patents.google.com/patent/US11816428B2/
Machine learning, artificial intelligence, and other computer-implemented methods are used to identify various semantically important chunks in documents, automatically label them with appropriate datatypes and semantic roles, and use this enhanced information to assist authors and to support downstream processes. Chunk locations, datatypes, and semantic roles can often be automatically determined from what is here called “context”, to wit, the combination of their formatting, structure, and content; those of adjacent or nearby content; overall patterns of occurrence in a document, and similarities of all these things across documents (mainly but not exclusively among documents in the same document set). Similarity is not limited to exact or fuzzy string or property comparisons, but may include similarity of natural language grammatical structure, ML (machine learning) techniques such as measuring similarity of word, chunk, and other embeddings, and the datatypes and semantic roles of previously-identified chunks.
@patent{Begun2023-in, author = {Begun, Andrew and DeRose, Steven and Jaffri, Taqi and Mart\'{i}, Luis and Palmer, Michael and Paoli, Jean and Pavlopoulou, Christina and Pricoiu, Elena and Sarangi, Swagatika and Sawicki, Martin and Shehadeh, Manar and Taron, Michael and Toprani, Bhaven and Wadia, Zubin Rustom and Watson, David and White, Eric and Fan, Joshua Yongshin and Gupta, Kush and Hoang, Andrew Minh and Liu, Zhanlin and Paliakkara, Jerome George and Wu, Zhaofeng and Zhang, Yue and Zhou, Xiaoquan}, holder = {Docugami, Inc.}, location = {US}, month = nov, number = {11816428}, title = {Automatically identifying chunks in sets of documents}, type = {patent}, url = {https://patents.google.com/patent/US11816428B2/}, year = {2023}, bdsk-url-1 = {https://patents.google.com/patent/US11816428B2/} } -
Begun, A., DeRose, S., Jaffri, T., Martí, L., Palmer, M., Paoli, J., Pavlopoulou, C., Pricoiu, E., Sarangi, S., Sawicki, M., Shehadeh, M., Taron, M., Toprani, B., Wadia, Z. R., Watson, D., White, E., Fan, J. Y., Gupta, K., Hoang, A. M., … Zhou, X. (2023). Enabling flexible processing of semantically-annotated documents (Patent) [Patent]. https://patents.google.com/patent/US11822880B2/
Machine learning, artificial intelligence, and other computer-implemented methods are used to identify various semantically important chunks in documents, automatically label them with appropriate datatypes and semantic roles, and use this enhanced information to assist authors and to support downstream processes. Chunk locations, datatypes, and semantic roles can often be automatically determined from what is here called “context”, to wit, the combination of their formatting, structure, and content; those of adjacent or nearby content; overall patterns of occurrence in a document, and similarities of all these things across documents (mainly but not exclusively among documents in the same document set). Similarity is not limited to exact or fuzzy string or property comparisons, but may include similarity of natural language grammatical structure, ML (machine learning) techniques such as measuring similarity of word, chunk, and other embeddings, and the datatypes and semantic roles of previously-identified chunks.
@patent{Begun2023-hk, author = {Begun, Andrew and DeRose, Steven and Jaffri, Taqi and Mart\'{i}, Luis and Palmer, Michael and Paoli, Jean and Pavlopoulou, Christina and Pricoiu, Elena and Sarangi, Swagatika and Sawicki, Martin and Shehadeh, Manar and Taron, Michael and Toprani, Bhaven and Wadia, Zubin Rustom and Watson, David and White, Eric and Fan, Joshua Yongshin and Gupta, Kush and Hoang, Andrew Minh and Liu, Zhanlin and Paliakkara, Jerome George and Wu, Zhaofeng and Zhang, Yue and Zhou, Xiaoquan}, holder = {Docugami, Inc.}, location = {US}, month = nov, number = {11822880}, title = {Enabling flexible processing of semantically-annotated documents}, type = {patent}, url = {https://patents.google.com/patent/US11822880B2/}, year = {2023}, bdsk-url-1 = {https://patents.google.com/patent/US11822880B2/} } -
Begun, A. P., DeRose, S., Jaffri, T., Martí, L., Palmer, M., Paoli, J., Pavlopoulou, C., Pricoiu, E., Sarangi, S., Sawicki, M., Shehadeh, M., Taron, M., Toprani, B., Wadia, Z. R., Watson, D., White, E., Fan, J. Y., Gupta, K., Hoang, A. M., … Zhou, X. (2022). Cross-document intelligent authoring and processing, including format for semantically-annotated documents (Patentus) [Patentus]. https://patents.google.com/patent/US11392763B2/
Machine learning, artificial intelligence, and other computer-implemented methods are used to identify various semantically important chunks in documents, automatically label them with appropriate datatypes and semantic roles, and use this enhanced information to assist authors and to support downstream processes. Chunk locations, datatypes, and semantic roles can often be automatically determined from what is here called “context”, to wit, the combination of their formatting, structure, and content; those of adjacent or nearby content; overall patterns of occurrence in a document, and similarities of all these things across documents (mainly but not exclusively among documents in the same document set). Similarity is not limited to exact or fuzzy string or property comparisons, but may include similarity of natural language grammatical structure, ML (machine learning) techniques such as measuring similarity of word, chunk, and other embeddings, and the datatypes and semantic roles of previously-identified chunks.
@patent{Begun2022-ps, author = {Begun, Andrew Paul and DeRose, Steven and Jaffri, Taqi and Mart\'{i}, Luis and Palmer, Michael and Paoli, Jean and Pavlopoulou, Christina and Pricoiu, Elena and Sarangi, Swagatika and Sawicki, Martin and Shehadeh, Manar and Taron, Michael and Toprani, Bhaven and Wadia, Zubin Rustom and Watson, David and White, Eric and Fan, Joshua Yongshin and Gupta, Kush and Hoang, Andrew Minh and Liu, Zhanlin and Paliakkara, Jerome George and Wu, Zhaofeng and Zhang, Yue and Zhou, Xiaoquan}, holder = {Docugami~Inc.}, location = {USA}, month = jul, number = {11392763}, title = {Cross-document intelligent authoring and processing, including format for semantically-annotated documents}, type = {patentus}, url = {https://patents.google.com/patent/US11392763B2/}, year = {2022}, bdsk-url-1 = {https://patents.google.com/patent/US11392763B2/} } -
Begun, A., DeRose, S., Jaffri, T., Martí, L., Palmer, M., Paoli, J., Pavlopoulou, C., Pricoiu, E., Sarangi, S., Sawicki, M., Shehadeh, M., Taron, M., Toprani, B., Wadia, Z. R., Watson, D., White, E., Fan, J. Y., Gupta, K., Hoang, A. M., … Zhou, X. (2022). Assisting authors via semantically-annotated documents (Patentus) [Patentus]. https://patents.google.com/patent/US11507740B2/
Machine learning, artificial intelligence, and other computer-implemented methods are used to identify various semantically important chunks in documents, automatically label them with appropriate datatypes and semantic roles, and use this enhanced information to assist authors and to support downstream processes. Chunk locations, datatypes, and semantic roles can often be automatically determined from what is here called “context”, to wit, the combination of their formatting, structure, and content; those of adjacent or nearby content; overall patterns of occurrence in a document, and similarities of all these things across documents (mainly but not exclusively among documents in the same document set). Similarity is not limited to exact or fuzzy string or property comparisons, but may include similarity of natural language grammatical structure, ML (machine learning) techniques such as measuring similarity of word, chunk, and other embeddings, and the datatypes and semantic roles of previously-identified chunks.
@patent{Begun2022-gd, author = {Begun, Andrew and DeRose, Steven and Jaffri, Taqi and Mart\'{i}, Luis and Palmer, Michael and Paoli, Jean and Pavlopoulou, Christina and Pricoiu, Elena and Sarangi, Swagatika and Sawicki, Martin and Shehadeh, Manar and Taron, Michael and Toprani, Bhaven and Wadia, Zubin Rustom and Watson, David and White, Eric and Fan, Joshua Yongshin and Gupta, Kush and Hoang, Andrew Minh and Liu, Zhanlin and Paliakkara, Jerome George and Wu, Zhaofeng and Zhang, Yue and Zhou, Xiaoquan}, holder = {Docugami, Inc.}, location = {USA}, month = nov, number = {11507740}, title = {Assisting authors via semantically-annotated documents}, type = {patentus}, url = {https://patents.google.com/patent/US11507740B2/}, year = {2022}, bdsk-url-1 = {https://patents.google.com/patent/US11507740B2/} } -
Begun, A. P., DeRose, S., Jaffri, T., Martí, L., Palmer, M., Paoli, J., Pavlopoulou, C., Pricoiu, E., Sarangi, S., Sawicki, M., Shehadeh, M., Taron, M., Toprani, B., Wadia, Z. R., Watson, D., White, E., Fan, J. Y., Gupta, K., Hoang, A. M., … Zhou, X. (2022). Automatically assigning semantic role labels to parts of documents (Patentus) [Patentus]. https://patents.google.com/patent/US11514238B2/
Machine learning, artificial intelligence, and other computer-implemented methods are used to identify various semantically important chunks in documents, automatically label them with appropriate datatypes and semantic roles, and use this enhanced information to assist authors and to support downstream processes. Chunk locations, datatypes, and semantic roles can often be automatically determined from what is here called “context”, to wit, the combination of their formatting, structure, and content; those of adjacent or nearby content; overall patterns of occurrence in a document, and similarities of all these things across documents (mainly but not exclusively among documents in the same document set). Similarity is not limited to exact or fuzzy string or property comparisons, but may include similarity of natural language grammatical structure, ML (machine learning) techniques such as measuring similarity of word, chunk, and other embeddings, and the datatypes and semantic roles of previously-identified chunks.
@patent{Begun2022-sc, author = {Begun, Andrew Paul and DeRose, Steven and Jaffri, Taqi and Mart\'{i}, Luis and Palmer, Michael and Paoli, Jean and Pavlopoulou, Christina and Pricoiu, Elena and Sarangi, Swagatika and Sawicki, Martin and Shehadeh, Manar and Taron, Michael and Toprani, Bhaven and Wadia, Zubin Rustom and Watson, David and White, Eric and Fan, Joshua Yongshin and Gupta, Kush and Hoang, Andrew Minh and Liu, Zhanlin and Paliakkara, Jerome George and Wu, Zhaofeng and Zhang, Yue and Zhou, Xiaoquan}, holder = {Docugami, Inc.}, location = {USA}, month = nov, number = {11514238}, title = {Automatically assigning semantic role labels to parts of documents}, type = {patentus}, url = {https://patents.google.com/patent/US11514238B2/}, year = {2022}, bdsk-url-1 = {https://patents.google.com/patent/US11514238B2/} } -
de Moraes, R. F., Evangelista, R. dos S., Fernandes, L. A. F., & Martí, L. (2021). GCOOD: A Generic Coupled Out-of-Distribution Detector for Robust Classification. 2021 34th SIBGRAPI Conference on Graphics, Patterns and Images (SIBGRAPI), 409–416. https://doi.org/10.1109/SIBGRAPI54419.2021.00062
Neural networks have achieved high degrees of accuracy in classification tasks. However, when an out-of-distribution (OOD) sample (i.e., entries from unknown classes) is submitted to the classification process, the result is the association of the sample to one or more of the trained classes with different degrees of confidence. If any of these confidence values are more significant than the user-defined threshold, the network will mislabel the sample, affecting the model credibility. The definition of the acceptance threshold itself is a sensitive issue in the face of the classifier’s overconfidence. This paper presents the Generic Coupled OOD Detector (GCOOD), a novel Convolutional Neural Network (CNN) tailored to detect whether an entry submitted to a trained classification model is an OOD sample for that model. From the analysis of the Softmax output of any classifier, our approach can indicate whether the resulting classification should be considered or not as a sample of some of the trained classes. To train our CNN, we had to develop a novel training strategy based on Voronoi diagrams of the location of representative entries in the latent space of the classification model and graph coloring. We evaluated our approach using ResNet, VGG, DenseNet, and SqueezeNet classifiers with images from the CIFAR-10 dataset.
@inproceedings{9643095, author = {de Moraes, Rog\'{e}rio Ferreira and Evangelista, Raphael dos S. and Fernandes, Leandro A. F. and Mart\'{i}, Luis}, booktitle = {2021 34th SIBGRAPI Conference on Graphics, Patterns and Images (SIBGRAPI)}, doi = {10.1109/SIBGRAPI54419.2021.00062}, hal_id = {hal-04396403}, hal_version = {v1}, issn = {2377-5416}, keywords = {Training ; Graphics ; Neural networks ; Detectors ; Convolutional neural networks ; Task analysis ; Faces}, location = {Gramado, Rio Grande do Sul, Brazil}, month = oct, pages = {409--416}, title = {{GCOOD}: {A} Generic Coupled Out-of-Distribution Detector for Robust Classification}, year = {2021}, bdsk-url-1 = {https://doi.org/10.1109/SIBGRAPI54419.2021.00062} } -
Santana, R., Martí, L., & Zhang, M. (2019). GP-based methods for domain adaptation: using brain decoding across subjects as a test-case. Genetic Programming and Evolvable Machines, 20(3). https://doi.org/10.1007/s10710-019-09352-6
Research on classifier transferability intends that the information gathered in the solution of a given classification problem could be reused in the solution of similar or related problems. We propose the evolution of transferable classifiers based on the use of multi-objective genetic programming and new fitness-functions that evaluate the amount of transferability. We focus on the domain adaptation scenario in which the problem to be solved is the same in the source and target domains, but the distribution of data is different between domains. As a real-world test case we address the brain decoding problem, whose goal is to predict the stimulus presented to a subject from the analysis of his brain activity. Brain decoding across subjects attempts to reuse the classifiers learned from some subjects in the classification of the others. We evolved GP-based classifiers using different variants of the introduced approach to test their effectiveness on data obtained from a brain decoding experiment involving 16 subjects. Our results show that the GP-based classifiers evolved trying to maximize transferability are able to improve classification accuracy over other classical classifiers that incorporate domain adaptation methods. Moreover, after comparing our algorithm to importance-weighted cross validation (in conjunction with many ML methods), we conclude that our approach achieves state of the art results in terms of transferability.
@article{Santana2019, author = {Santana, Roberto and Mart\'{i}, Luis and Zhang, Mengjie}, day = {10}, doi = {10.1007/s10710-019-09352-6}, issn = {1573-7632}, journal = {Genetic Programming and Evolvable Machines}, month = may, number = {3}, title = {{GP}-based methods for domain adaptation: using brain decoding across subjects as a test-case}, url = {https://doi.org/10.1007/s10710-019-09352-6}, volume = {20}, year = {2019}, bdsk-url-1 = {https://doi.org/10.1007/s10710-019-09352-6} } -
Begun, A. P., DeRose, S., Jaffri, T., Martí, L., Palmer, M. B., Paoli, J., Pavlopoulou, C., Pricoiu, E., Sarangi, S., Sawicki, M., Shehadeh, M., Taron, M., Toprani, B., Wadia, Z. R., Watson, D., White, E., Fan, J. Y., Gupta, K., Hoang, A. M., … Zhou, X. (2019). Cross-document intelligent writing and processing assistant (Patreq) [Patreq]. https://patents.google.com/patent/EP4028961A4
Machine learning, artificial intelligence, and other computer-implemented methods are used to identify various semantically important chunks in documents, automatically label them with appropriate datatypes and semantic roles, and use this enhanced information to assist authors and to support downstream processes. Chunk locations, datatypes, and semantic roles can often be automatically determined from what is here called “context”, to wit, the combination of their formatting, structure, and content; those of adjacent or nearby content; overall patterns of occurrence in a document, and similarities of all these things across documents (mainly but not exclusively among documents in the same document set). Similarity is not limited to exact or fuzzy string or property comparisons, but may include similarity of natural language grammatical structure, ML (machine learning) techniques such as measuring similarity of word, chunk, and other embeddings, and the datatypes and semantic roles of previously-identified chunks..
@patent{A2022-ca, author = {Begun, Andrew Paul and DeRose, Steven and Jaffri, Taqi and Mart\'{i}, Luis and Palmer, Michael B and Paoli, Jean and Pavlopoulou, Christina and Pricoiu, Elena and Sarangi, Swagatika and Sawicki, Martin and Shehadeh, Manar and Taron, Michael and Toprani, Bhaven and Wadia, Zubin Rustom and Watson, David and White, Eric and Fan, Joshua Yongshin and Gupta, Kush and Hoang, Andrew Minh and Liu, Zhanlin and Paliakkara, Jerome George and Wu, Zhaofeng and Zhang, Yue and Zhou, Xiaoquan}, holder = {Docugami~Inc.}, keywords = {pending}, location = {Canada}, month = sep, note = {pending}, number = {CA3150535A1}, title = {Cross-document intelligent writing and processing assistant}, type = {patreq}, url = {https://patents.google.com/patent/EP4028961A4}, year = {2019}, bdsk-url-1 = {https://patents.google.com/patent/EP4028961A4} } -
Begun, A. P., DeRose, S., Jaffri, T., Martí, L., Palmer, M. B., Paoli, J., Pavlopoulou, C., Pricoiu, E., Sarangi, S., Sawicki, M., Shehadeh, M., Taron, M., Toprani, B., Wadia, Z. R., Watson, D., White, E., Fan, J. Y., Gupta, K., Hoang, A. M., … Zhou, X. (2019). Cross-document intelligent writing and processing assistant (Patreq) [Patreq]. https://patents.google.com/patent/EP4028961A4
Machine learning, artificial intelligence, and other computer-implemented methods are used to identify various semantically important blocks in a document, automatically mark them with the appropriate data types and semantic roles, and use this enhanced information to assist authors and support downstream processes. Block location, data type, and semantic role can often be automatically determined from: the “contexts” referred to herein, i.e., the combination of their formatting, structure, and content; those of adjacent or nearby content; global appearance patterns in the document; and the similarity of all these things across documents (primarily but not exclusively among documents in the same document set). Similarity is not limited to exact or fuzzy character strings or attribute comparisons, but may include similarity of natural language grammar structures, ML (machine learning) techniques such as measured similarity of words, blocks, and other embeddings, and similarity of data types and semantic roles of previously identified blocks.
@patent{A2022-cn, author = {Begun, Andrew Paul and DeRose, Steven and Jaffri, Taqi and Mart\'{i}, Luis and Palmer, Michael B and Paoli, Jean and Pavlopoulou, Christina and Pricoiu, Elena and Sarangi, Swagatika and Sawicki, Martin and Shehadeh, Manar and Taron, Michael and Toprani, Bhaven and Wadia, Zubin Rustom and Watson, David and White, Eric and Fan, Joshua Yongshin and Gupta, Kush and Hoang, Andrew Minh and Liu, Zhanlin and Paliakkara, Jerome George and Wu, Zhaofeng and Zhang, Yue and Zhou, Xiaoquan}, holder = {Docugami~Inc.}, keywords = {pending}, location = {China}, month = sep, note = {pending}, number = {114616572:A}, title = {Cross-document intelligent writing and processing assistant}, type = {patreq}, url = {https://patents.google.com/patent/EP4028961A4}, year = {2019}, bdsk-url-1 = {https://patents.google.com/patent/EP4028961A4} } -
Begun, A. P., DeRose, S., Jaffri, T., Martí, L., Palmer, M. B., Paoli, J., Pavlopoulou, C., Pricoiu, E., Sarangi, S., Sawicki, M., Shehadeh, M., Taron, M., Toprani, B., Wadia, Z. R., Watson, D., White, E., Fan, J. Y., Gupta, K., Hoang, A. M., … Zhou, X. (2019). Cross-document intelligent writing and processing assistant (Patreq) [Patreq]. https://patents.google.com/patent/EP4028961A4
Machine learning, artificial intelligence, and other computer-implemented methods are used to identify various semantically important blocks in a document, automatically mark them with the appropriate data types and semantic roles, and use this enhanced information to assist authors and support downstream processes. Block location, data type, and semantic role can often be automatically determined from: the “contexts” referred to herein, i.e., the combination of their formatting, structure, and content; those of adjacent or nearby content; global appearance patterns in the document; and the similarity of all these things across documents (primarily but not exclusively among documents in the same document set). Similarity is not limited to exact or fuzzy character strings or attribute comparisons, but may include similarity of natural language grammar structures, ML (machine learning) techniques such as measured similarity of words, blocks, and other embeddings, and similarity of data types and semantic roles of previously identified blocks.
@patent{A2022-jp, author = {Begun, Andrew Paul and DeRose, Steven and Jaffri, Taqi and Mart\'{i}, Luis and Palmer, Michael B and Paoli, Jean and Pavlopoulou, Christina and Pricoiu, Elena and Sarangi, Swagatika and Sawicki, Martin and Shehadeh, Manar and Taron, Michael and Toprani, Bhaven and Wadia, Zubin Rustom and Watson, David and White, Eric and Fan, Joshua Yongshin and Gupta, Kush and Hoang, Andrew Minh and Liu, Zhanlin and Paliakkara, Jerome George and Wu, Zhaofeng and Zhang, Yue and Zhou, Xiaoquan}, holder = {Docugami~Inc.}, keywords = {pending}, location = {Japan}, month = sep, note = {pending}, number = {JP2022547750A}, title = {Cross-document intelligent writing and processing assistant}, type = {patreq}, url = {https://patents.google.com/patent/EP4028961A4}, year = {2019}, bdsk-url-1 = {https://patents.google.com/patent/EP4028961A4} } -
Begun, A. P., DeRose, S., Jaffri, T., Martí, L., Palmer, M. B., Paoli, J., Pavlopoulou, C., Pricoiu, E., Sarangi, S., Sawicki, M., Shehadeh, M., Taron, M., Toprani, B., Wadia, Z. R., Watson, D., White, E., Fan, J. Y., Gupta, K., Hoang, A. M., … Zhou, X. (2019). Cross-document intelligent writing and processing assistant (Patreqeu) [Patreqeu]. https://patents.google.com/patent/EP4028961A4
Machine learning, artificial intelligence, and other computer-implemented methods are used to identify various semantically important blocks in a document, automatically mark them with the appropriate data types and semantic roles, and use this enhanced information to assist authors and support downstream processes. Block location, data type, and semantic role can often be automatically determined from: the “contexts” referred to herein, i.e., the combination of their formatting, structure, and content; those of adjacent or nearby content; global appearance patterns in the document; and the similarity of all these things across documents (primarily but not exclusively among documents in the same document set). Similarity is not limited to exact or fuzzy character strings or attribute comparisons, but may include similarity of natural language grammar structures, ML (machine learning) techniques such as measured similarity of words, blocks, and other embeddings, and similarity of data types and semantic roles of previously identified blocks.
@patent{A2022-eu, author = {Begun, Andrew Paul and DeRose, Steven and Jaffri, Taqi and Mart\'{i}, Luis and Palmer, Michael B and Paoli, Jean and Pavlopoulou, Christina and Pricoiu, Elena and Sarangi, Swagatika and Sawicki, Martin and Shehadeh, Manar and Taron, Michael and Toprani, Bhaven and Wadia, Zubin Rustom and Watson, David and White, Eric and Fan, Joshua Yongshin and Gupta, Kush and Hoang, Andrew Minh and Liu, Zhanlin and Paliakkara, Jerome George and Wu, Zhaofeng and Zhang, Yue and Zhou, Xiaoquan}, holder = {Docugami~Inc.}, keywords = {pending}, location = {EU}, month = sep, note = {pending}, number = {EP4028961A4}, title = {Cross-document intelligent writing and processing assistant}, type = {patreqeu}, url = {https://patents.google.com/patent/EP4028961A4}, year = {2019}, bdsk-url-1 = {https://patents.google.com/patent/EP4028961A4} } -
Begun, A. P., DeRose, S., Jaffri, T., Martí, L., Palmer, M. B., Paoli, J., Pavlopoulou, C., Pricoiu, E., Sarangi, S., Sawicki, M., Shehadeh, M., Taron, M., Toprani, B., Wadia, Z. R., Watson, D., White, E., Fan, J. Y., Gupta, K., Hoang, A. M., … Zhou, X. (2019). Cross-document intelligent writing and processing assistant (Patreq) [Patreq]. https://patents.google.com/patent/EP4028961A4
Machine learning, artificial intelligence, and other computer-implemented methods are used to identify various semantically important blocks in a document, automatically mark them with the appropriate data types and semantic roles, and use this enhanced information to assist authors and support downstream processes. Block location, data type, and semantic role can often be automatically determined from: the “contexts” referred to herein, i.e., the combination of their formatting, structure, and content; those of adjacent or nearby content; global appearance patterns in the document; and the similarity of all these things across documents (primarily but not exclusively among documents in the same document set). Similarity is not limited to exact or fuzzy character strings or attribute comparisons, but may include similarity of natural language grammar structures, ML (machine learning) techniques such as measured similarity of words, blocks, and other embeddings, and similarity of data types and semantic roles of previously identified blocks.
@patent{A2022-kr, author = {Begun, Andrew Paul and DeRose, Steven and Jaffri, Taqi and Mart\'{i}, Luis and Palmer, Michael B and Paoli, Jean and Pavlopoulou, Christina and Pricoiu, Elena and Sarangi, Swagatika and Sawicki, Martin and Shehadeh, Manar and Taron, Michael and Toprani, Bhaven and Wadia, Zubin Rustom and Watson, David and White, Eric and Fan, Joshua Yongshin and Gupta, Kush and Hoang, Andrew Minh and Liu, Zhanlin and Paliakkara, Jerome George and Wu, Zhaofeng and Zhang, Yue and Zhou, Xiaoquan}, holder = {Docugami~Inc.}, keywords = {pending,}, location = {Korea}, month = sep, note = {pending}, number = {10-2022-7011501}, title = {Cross-document intelligent writing and processing assistant}, type = {patreq}, url = {https://patents.google.com/patent/EP4028961A4}, year = {2019}, bdsk-url-1 = {https://patents.google.com/patent/EP4028961A4} } -
Martí, L., & Schoenauer, M. (2018). Bio-inspired Approaches to Anomaly and Intrusion Detection. Proceedings of the Genetic and Evolutionary Computation Conference Companion, 1121–1137. https://doi.org/10.1145/3205651.3207853
@inproceedings{Marti:2018:BAA:3205651.3207853, acmid = {3207853}, keywords = {talk}, address = {New York, NY, USA}, author = {Mart\'{i}, Luis and Schoenauer, Marc}, bdsk-url-2 = {https://doi.org/10.1145/3205651.3207853}, booktitle = {Proceedings of the Genetic and Evolutionary Computation Conference Companion}, doi = {10.1145/3205651.3207853}, isbn = {978-1-4503-5764-7}, location = {Kyoto, Japan}, numpages = {17}, pages = {1121--1137}, publisher = {ACM}, series = {GECCO'18}, title = {Bio-inspired Approaches to Anomaly and Intrusion Detection}, url = {http://doi.acm.org/10.1145/3205651.3207853}, year = {2018}, bdsk-url-1 = {http://doi.acm.org/10.1145/3205651.3207853} } -
Pereira, C. S., Dias, D. M., Vellasco, M. M. B. R., Viana, F. H. F., & Martí, L. (2018). Crude Oil Refinery Scheduling: Addressing a Real-world Multiobjective Problem Through Genetic Programming and Dominance-based Approaches. Proceedings of the Genetic and Evolutionary Computation Conference Companion, 1821–1828. https://doi.org/10.1145/3205651.3208291
This study presents the crude oil scheduling problem with four objectives divided in two different levels of importance. It comes from a real refinery where the scheduling starts on the offloading of ships, encompasses terminal and refinery tanks, a crude pipeline, and finishes on the output streams of the crude distillation units. We propose a new approach for the Quantum-Inspired Grammar-based Linear Genetic Programming (QIGLGP) evolutionary algorithm to handle the multiple objectives of the problem using the non-dominance concept. The modifications are concentrated on the population updating and sorting steps of QIGLGP. We tackle difference of importance among the objectives using the principle of violation of constraints. The problem constraints define if an instruction will or not be executed but do not affect the violation equation of the objectives. The individuals which have objective values under a pre-defined upper limit are better ranked. Results from five scenarios showed that the proposed model was able to significantly increase the percentage of runs with acceptable solutions, achieving success ratio of 100% in 3 cases and over 70% in 2 other ones. They also show that the Pareto front of these accepted runs contains a set of non-dominated solutions that could be analyzed by the decision maker for his a posteriori decision.
@inproceedings{pereira-2018:scheduling, acmid = {3208291}, address = {New York, NY, USA}, author = {Pereira, Cristiane Salgado and Dias, Douglas Motta and Vellasco, Marley M. B. R. and Viana, Francisco Henrique F and Mart\'{i}, Luis}, booktitle = {Proceedings of the Genetic and Evolutionary Computation Conference Companion}, doi = {10.1145/3205651.3208291}, isbn = {978-1-4503-5764-7}, keywords = {crude oil refinery scheduling, evolutionary multiobjective optimization algorithm, quantum-inspired genetic programming}, location = {Kyoto, Japan}, numpages = {8}, pages = {1821--1828}, publisher = {ACM}, series = {GECCO '18}, title = {Crude Oil Refinery Scheduling: {A}ddressing a Real-world Multiobjective Problem Through {G}enetic {P}rogramming and Dominance-based Approaches}, url = {http://doi.acm.org/10.1145/3205651.3208291}, year = {2018}, bdsk-url-1 = {http://doi.acm.org/10.1145/3205651.3208291}, bdsk-url-2 = {https://doi.org/10.1145/3205651.3208291} } -
Ferreira de Moraes, R., Villela da Silva, A., Satoru Ochi, L., & Martí, L. (2018). Implementation of a RVND, VNS, ILS heuristic for the Traveling Car Renter Problem. Proceedings of the 2018 IEEE Congress on Evolutionary Computation (IEEE CEC 2018) Part of the 2018 IEEE World Congress on Computational Intelligence (IEEE WCCI 2018), 1–8. https://doi.org/10.1109/CEC.2018.8477777
This work proposes a new algorithm that combines Iterated Local Search (ILS), Variable Neighborhood Search (VNS), Random Variable Neighborhood Descent (RVND) and a constructive initial search to solve the Traveling Car Renter Problem (CaRS). The CaRS a variation of the Traveling Salesman Problem where a customer wants to visit a set of cities using a rental car. The customer has a fixed amount of car options available to choose and when he returns the car, he will not be able to use it again. The traveling cost varies according to the car used, being the same value in the opposite direction. In addition to this cost, the customer also pays for a vehicle return. The objective is to minimize costs and fees. The results were compared to the state of the art and showed better performances.
@inproceedings{rogerio-2018:cec, author = {Ferreira de Moraes, Rog\'{e}rio and Villela da Silva, Andr\'{e} and Satoru Ochi, Luiz and Mart\'{i}, Luis}, booktitle = {Proceedings of the 2018 IEEE Congress on Evolutionary Computation (IEEE CEC 2018) part of the 2018 IEEE World Congress on Computational Intelligence (IEEE WCCI 2018)}, doi = {10.1109/CEC.2018.8477777}, keywords = {iterative methods;search problems;transportation;travelling salesman problems;traveling cost varies;constructive initial search;CaRS;traveling car renter problem;random variable neighborhood descent;traveling salesman problem;RVND;VNS;ILS;iterated local search;variable neighborhood search;Automobiles;Urban areas;Matrix decomposition;Search problems;Traveling salesman problems;Electronic mail;Random variables}, location = {Rio de Janeiro, Brazil}, month = jul, pages = {1--8}, publisher = {IEEE Press}, title = {Implementation of a {RVND}, {VNS}, {ILS} heuristic for the Traveling Car Renter Problem}, year = {2018}, bdsk-url-1 = {https://doi.org/10.1109/CEC.2018.8477777} } -
Martí, L., Fansi-Tchango, A., Navarro, L., & Schoenauer, M. (2017). Progressively Adding Objectives: A Case Study in Anomaly Detection. Proceedings of the 2017 Annual Conference on Genetic and Evolutionary Computation (GECCO’17), 1–8. https://doi.org/10.1145/3071178.3071333
One of the principles of evolutionary multi-objective optimization is the conjoint optimization of the objective functions. However, in some cases, some of the objectives are easier to attain than others. This causes the population to lose diversity at a high rate and stagnate in early stages of the evolution. This paper presents the progressive addition of objectives (PAO) heuristic. PAO gradually adds objectives to a given problem relying on a perceived measure of complexity. This diversity loss phenomenon caused by the nature of a given objective has been observed when applying the Voronoi diagram-based evolutionary algorithm (VorEAl) in anomaly detection problems. Consequently, PAO has been first directed to address that issue. The experimental studies carried out show that the PAO heuristic manages to yield better results than the direct use of VorEAl on a group of test problems.
@inproceedings{marti-2017:gecco, address = {New York, NY, USA}, author = {Mart\'{i}, Luis and Fansi-Tchango, Arsene and Navarro, Laurent and Schoenauer, Marc}, booktitle = {Proceedings of the 2017 Annual Conference on Genetic and Evolutionary Computation (GECCO'17)}, doi = {10.1145/3071178.3071333}, keywords = {anomaly detection, artificial immune systems, multi-objective optimization, voronoi diagrams}, location = {Berlin, Germany}, pages = {1--8}, publisher = {ACM Press}, title = {Progressively Adding Objectives: {A} Case Study in Anomaly Detection}, year = {2017}, bdsk-url-1 = {https://doi.org/10.1145/3071178.3071333} } -
Marinho, W., & Martí, L. (2016). Review of Student Proficiency Modeling Techniques for use in Intelligent Tutoring Systems. Workshop De Pesquisa e Desenvolvimento Em Inteligêcia Artificial, Inteligêcia Collectiva e Ciêcia De Dados. http://www.addlabs.uff.br/workpedia2016/anais-do-workpedia-2016/
Abstract not available at the time of preparation of this document.
@inproceedings{marinho-2016:workpedia, author = {Marinho, Wemerson and Mart\'{i}, Luis}, booktitle = {Workshop de Pesquisa e Desenvolvimento em Intelig\^{e}cia Artificial, Intelig\^{e}cia Collectiva e Ci\^{e}cia de Dados}, title = {Review of Student Proficiency Modeling Techniques for use in Intelligent Tutoring Systems}, url = {http://www.addlabs.uff.br/workpedia2016/anais-do-workpedia-2016/}, year = {2016}, bdsk-url-1 = {http://www.addlabs.uff.br/workpedia2016/anais-do-workpedia-2016/} } -
Dias de Mello Jr, H., Martí, L., Abs da Cruz, A. V., & Rebuzzi Vellasco, M. M. B. (2016). Evolutionary algorithms and elliptical copulas applied to continuous optimization problems. Information Sciences, 369, 419–440. https://doi.org/10.1016/j.ins.2016.07.006
Abstract Estimation of Distribution Algorithms (EDAs) constitutes a class of evolutionary algorithms that can extract and exploit knowledge acquired throughout the optimization process. The most critical step in the EDAs is the estimation of the joint probability distribution associated to the variables from the most promising solutions determined by the evaluation function. Recently, a new approach to EDAs has been developed, based on copula theory, to improve the estimation of the joint probability distribution function. However, most copula-based EDAs still present two major drawbacks: focus on copulas with constant parameters, and premature convergence. This paper presents a new copula-based estimation of distribution algorithm for numerical optimization problems, named EDA based on Multivariate Elliptical Copulas (EDA-MEC). This model uses multivariate copulas to estimate the probability distribution for generating a population of individuals. The EDA-MEC differs from other copula-based EDAs in several aspects: the copula parameter is dynamically estimated, using dependence measures; it uses a variation of the learned probability distribution to generate individuals that help to avoid premature convergence; and uses a heuristic to reinitialize the population as an additional technique to preserve the diversity of solutions. The paper shows, by means of a set of parametric tests, that this approach improves the overall performance of the optimization process when compared with other copula-based EDAs and with other efficient heuristics such as the Covariance Matrix Adaptation Evolution Strategy (CMA-ES).
@article{harold-2016:copulas, author = {Dias de Mello Jr, Harold and Mart\'{i}, Luis and Abs da Cruz, Andr\'{e} V. and Rebuzzi Vellasco, Marley~M.~B.}, doi = {10.1016/j.ins.2016.07.006}, issn = {0020-0255}, journal = {Information Sciences}, month = nov, pages = {419--440}, title = {Evolutionary algorithms and elliptical copulas applied to continuous optimization problems}, volume = {369}, year = {2016}, bdsk-url-1 = {https://doi.org/10.1016/j.ins.2016.07.006} } -
Martí, L., Fansi-Tchango, A., Navarro, L., & Schoenauer, M. (2016). VorAIS: A Multi-Objective Voronoi Diagram-based Artificial Immune System. Proceedings of the 2016 Annual Conference on Genetic and Evolutionary Computation (GECCO’16), 11–12. https://doi.org/10.1145/2908961.2909027
This paper introduces the Voronoi diagram-based Artificial Immune System (VorAIS). VorAIS models the self/non-self using a Voronoi diagram that determines which areas of the problem domain correspond to self or to non-self. The diagram is evolved using a multi-objective bio-inspired approach in order to conjointly optimize various classification metrics (accuracy, recall and specificity). VorAIS is experimentally validated, first on standard classification problems, then on the well-known NSL-KDD dataset for anomaly detection where it favorably compares with other AIS approaches.
@inproceedings{marti-2016:gecco, address = {New York, NY, USA}, author = {Mart\'{i}, Luis and Fansi-Tchango, Arsene and Navarro, Laurent and Schoenauer, Marc}, booktitle = {Proceedings of the 2016 Annual Conference on Genetic and Evolutionary Computation (GECCO'16)}, doi = {10.1145/2908961.2909027}, isbn = {978-1-4503-4323-7}, keywords = {anomaly detection, artificial immune systems, multi-objectie optimization, Voronoi diagrams}, location = {Denver (CO) USA}, pages = {11--12}, publisher = {ACM Press}, title = {{VorAIS}: {A} Multi-Objective {V}oronoi Diagram-based Artificial Immune System}, year = {2016}, bdsk-url-1 = {https://doi.org/10.1145/2908961.2909027} } -
Martí, L., Fansi-Tchango, A., Navarro, L., & Schoenauer, M. (2016). Anomaly Detection with the Voronoi Diagram Evolutionary Algorithm. In J. Handl, E. Hart, R. P. Lewis, M. López-Ibáñez, G. Ochoa, & B. Paechter (Eds.), Proceedings of the 14th International Conference Parallel Problem Solving from Nature (PPSN XIV) (pp. 697–706). Springer International Publishing. https://doi.org/10.1007/978-3-319-45823-6_65
This paper presents the Voronoi diagram-based evolutionary algorithm (VorEAl). VorEAl partitions input space in abnormal/normal subsets using Voronoi diagrams. Diagrams are evolved using a multi-objective bio-inspired approach in order to conjointly optimize classification metrics while also being able to represent areas of the data space that are not present in the training dataset. As part of the paper, VorEAl is experimentally validated and contrasted with similar approaches.
@inproceedings{marti-2016:ppsn, author = {Mart\'{i}, Luis and Fansi-Tchango, Arsene and Navarro, Laurent and Schoenauer, Marc}, booktitle = {Proceedings of the 14th International Conference Parallel Problem Solving from Nature (PPSN XIV)}, doi = {10.1007/978-3-319-45823-6_65}, editor = {Handl, Julia and Hart, Emma and Lewis, R. Peter and L\'{o}pez-Ib\'{a}\~{n}ez, Manuel and Ochoa, Gabriela and Paechter, Ben}, isbn = {978-3-319-45823-6}, location = {Edinburgh, UK}, month = sep, pages = {697--706}, publisher = {Springer International Publishing}, title = {Anomaly Detection with the {V}oronoi Diagram Evolutionary Algorithm}, year = {2016}, bdsk-url-1 = {https://doi.org/10.1007/978-3-319-45823-6_65} } -
Martí, L., García, J., Berlanga, A., & Molina López, J. M. (2016). A Stopping Criterion for Multi-Objective Optimization Evolutionary Algorithms. Information Sciences, 367–368, 700–718. https://doi.org/10.1016/j.ins.2016.07.025
Abstract This paper puts forward a comprehensive study of the design of global stopping criteria for multi-objective optimization. In this study we propose a global stopping criterion, which is terms as MGBM after the authors surnames. MGBM combines a novel progress indicator, called mutual domination rate (MDR) indicator, with a simplified Kalman filter, which is used for evidence-gathering purposes. The MDR indicator, which is also introduced, is a special-purpose progress indicator designed for the purpose of stopping a multi-objective optimization. As part of the paper we describe the criterion from a theoretical perspective and examine its performance on a number of test problems. We also compare this method with similar approaches to the issue. The results of these experiments suggest that MGBM is a valid and accurate approach.
@article{marti-2016:stopping, author = {Mart\'{i}, Luis and Garc\'{i}a, Jes\'{u}s and Berlanga, Antonio and Molina L\'{o}pez, Jos\'{e} Manuel}, bdsk-url-2 = {https://doi.org/10.1016/j.ins.2016.07.025}, doi = {10.1016/j.ins.2016.07.025}, issn = {0020-0255}, journal = {Information Sciences}, month = jul, pages = {700--718}, title = {A Stopping Criterion for Multi-Objective Optimization Evolutionary Algorithms}, url = {http://www.sciencedirect.com/science/article/pii/S0020025516305072}, volume = {367--368}, year = {2016}, bdsk-url-1 = {http://www.sciencedirect.com/science/article/pii/S0020025516305072} } -
Brockhoff, D., Ehrgott, M., Figueira, J. R., Martí, L., Paquete, L., Stiglmayr, M., & Vanderpooten, D. (2015). Computational Complexity (WG2). In S. Greco, K. Klamroth, J. D. Knowles, & G. Rudolph (Eds.), Understanding Complexity in Multiobjective Optimization — Report from Dagstuhl Seminar 15031 (pp. 116–121). Schloss Dagstuhl, Leibniz-Zentrum für Informatik, Dagstuhl Publishing. http://drops.dagstuhl.de/opus/volltexte/2015/5037
This report documents the program and outcomes of the Dagstuhl Seminar 15031 Understanding Complexity in Multiobjective Optimization. This seminar carried on the series of four previous Dagstuhl Seminars (04461, 06501, 09041 and 12041) that were focused on Multiobjective Optimization, and strengthening the links between the Evolutionary Multiobjective Optimization (EMO) and Multiple Criteria Decision Making (MCDM) communities. The purpose of the seminar was to bring together researchers from the two communities to take part in a wide-ranging discussion about the different sources and impacts of complexity in multiobjective optimization. The outcome was a clarified viewpoint of complexity in the various facets of multiobjective optimization, leading to several research initiatives with innovative approaches for coping with complexity.
@inproceedings{brockhoff2015computational, address = {Dagstuhl, Germany}, author = {Brockhoff, Dimo and Ehrgott, Matthias and Figueira, Jos\'{e} Rui and Mart\'{i}, Luis and Paquete, Lu\'{i}s and Stiglmayr, Michael and Vanderpooten, Daniel}, booktitle = {Understanding Complexity in Multiobjective Optimization --- Report from Dagstuhl Seminar 15031}, editor = {Greco, Salvatore and Klamroth, Kathrin and Knowles, Joshua D. and Rudolph, G\"unter}, pages = {116--121}, publisher = {Schloss Dagstuhl, Leibniz-Zentrum f\"ur Informatik, Dagstuhl Publishing}, title = {Computational Complexity ({WG2})}, url = {http://drops.dagstuhl.de/opus/volltexte/2015/5037}, year = {2015}, bdsk-url-1 = {http://drops.dagstuhl.de/opus/volltexte/2015/5037} } -
Martí, L., Grimme, C., Kerschke, P., Trautmann, H., & Rudolph, G. (2015). Averaged Hausdorff Approximations of Pareto Fronts Based on Multiobjective Estimation of Distribution Algorithms. In J. L. Jiménez Laredo, S. Silva, & A. I. Esparcia-Alcázar (Eds.), Proceedings of the Companion Publication of the 2015 Annual Conference on Genetic and Evolutionary Computation (pp. 1427–1428). ACM Press. https://doi.org/10.1145/2739482.2764631
We propose a post-processing strategy which consists of applying the averaged Hausdorff indicator to the complete archive of solutions after optimization by multiobjective estimation of distribution algorithms (MEDAs) to select a uniformly distributed subset of non-dominated solutions.
@inproceedings{marti-2015:hausdorff, acmid = {2764631}, address = {New York, NY, USA}, author = {Mart\'{i}, Luis and Grimme, Christian and Kerschke, Paskal and Trautmann, Heike and Rudolph, G\"{u}nter}, booktitle = {Proceedings of the Companion Publication of the 2015 Annual Conference on Genetic and Evolutionary Computation}, doi = {10.1145/2739482.2764631}, editor = {Jim\'{e}nez Laredo, Juan Luis and Silva, Sara and Esparcia{-}Alc\'{a}zar, Anna Isabel}, isbn = {978-1-4503-3488-4}, keywords = {averaged hausdorff distance, estimation of distribution algorithm, multiobjective optimization}, location = {Madrid, Spain}, pages = {1427--1428}, publisher = {ACM Press}, series = {GECCO Companion '15}, title = {Averaged {H}ausdorff Approximations of {P}areto Fronts Based on Multiobjective Estimation of Distribution Algorithms}, year = {2015}, bdsk-url-1 = {https://doi.org/10.1145/2739482.2764631} } -
Martí, L., García, J., Berlanga, A., & Molina López, J. M. (2013). Multi-Objective Optimization with an Adaptive Resonance Theory-based Estimation of Distribution Algorithm. Annals of Mathematics and Artificial Intelligence, 68(4), 247–273. https://doi.org/10.1007/s10472-012-9303-0
The introduction of learning to the search mechanisms of optimization algorithms has been nominated as one of the viable approaches when dealing with complex optimization problems, in particular with multi-objective ones. One of the forms of carrying out this hybridization process is by using multi-objective optimization estimation of distribution algorithms (MOEDAs). However, it has been pointed out that current MOEDAs have an intrinsic shortcoming in their model-building algorithms that hamper their performance. In this work, we put forward the argument that error-based learning, the class of learning most commonly used in MOEDAs is responsible for current MOEDA underachievement. We present adaptive resonance theory (ART) as a suitable learning paradigm alternative and present a novel algorithm called multi-objective ART-based EDA (MARTEDA) that uses a Gaussian ART neural network for model-building and a hypervolume-based selector as described for the HypE algorithm. In order to assert the improvement obtained by combining two cutting-edge approaches to optimization an extensive set of experiments are carried out. These experiments also test the scalability of MARTEDA as the number of objective functions increases.
@article{marti-2012:marteda-amai, acmid = {2560185}, address = {Hingham, MA, USA}, author = {Mart\'{i}, Luis and Garc\'{i}a, Jes\'{u}s and Berlanga, Antonio and Molina L\'{o}pez, Jos\'{e} Manuel}, doi = {10.1007/s10472-012-9303-0}, issn = {1012-2443}, journal = {Annals of Mathematics and Artificial Intelligence}, keywords = {65K10, 68T05, 68T20, Adaptive resonance theory, Estimation of distribution algorithms, Multi-objective optimization}, number = {4}, numpages = {27}, pages = {247--273}, publisher = {Kluwer Academic Publishers}, title = {Multi-Objective Optimization with an Adaptive Resonance Theory-based Estimation of Distribution Algorithm}, volume = {68}, year = {2013}, bdsk-url-1 = {https://doi.org/10.1007/s10472-012-9303-0} } -
Martí, L., García, J., Berlanga, A., & Molina López, J. M. (2011). Indicator-based MONEDA: A Comparative Study of Scalability with Respect to Decision Space Dimensions. 2011 IEEE Conference on Evolutionary Computation (CEC), 957–964. https://doi.org/10.1109/CEC.2011.5949721
The multi-objective neural EDA (MONEDA) was proposed with the aim of overcoming some difficulties of current MOEDAs. MONEDA has been shown to yield relevant results when confronted with complex problems. Furthermore, its performance has been shown to adequately adapt to problems with many objectives. Nevertheless, one key issue remains to be studied: MONEDA scalability with regard to the number of decision variables. This paper has a two-fold purpose. On one hand we propose a modification of MONEDA that incorporates an indicator-based selection mechanism based on the HypE algorithm, while, on the other, we assess the indicator-based MONEDA when solving some complex two-objective problems, in particular problems UF1 to UF7 of the CEC 2009 MOP competition, configured with a progressively-increasing number of decision variables.
@inproceedings{marti-2011:indicator-moneda, address = {Piscataway, New Jersey}, author = {Mart\'{i}, Luis and Garc\'{i}a, Jes\'{u}s and Berlanga, Antonio and Molina L\'{o}pez, Jos\'{e} Manuel}, booktitle = {2011 IEEE Conference on Evolutionary Computation (CEC)}, doi = {10.1109/CEC.2011.5949721}, location = {New Orleans, Louisiana}, pages = {957--964}, publisher = {IEEE Press}, title = {Indicator-based {MONEDA}: {A} Comparative Study of Scalability with Respect to Decision Space Dimensions}, year = {2011}, bdsk-url-1 = {https://doi.org/10.1109/CEC.2011.5949721} } -
Martí, L., García, J., Berlanga, A., & Molina López, J. M. (2011). Multi-Objective Optimization with an Adaptive Resonance Theory-Based Estimation of Distribution Algorithm: A Comparative Study. In C. A. Coello Coello (Ed.), Learning and Intelligent Optimization (Vol. 6683, pp. 458–472). Springer. https://doi.org/10.1007/978-3-642-25566-3_36
The introduction of learning to the search mechanisms of optimization algorithms has been nominated as one of the viable approaches when dealing with complex optimization problems, in particular with multi-objective ones. One of the forms of carrying out this hybridization process is by using multi-objective optimization estimation of distribution algorithms (MOEDAs). However, it has been pointed out that current MOEDAs have a intrinsic shortcoming in their model-building algorithms that hamper their performance. In this work we argue that error-based learning, the class of learning most commonly used in MOEDAs is responsible for current MOEDA underachievement. We present adaptive resonance theory (ART) as a suitable learning paradigm alternative and present a novel algorithm called multi-objective ART-based EDA (MARTEDA) that uses a Gaussian ART neural network for model-building and an hypervolume-based selector as described for the HypE algorithm. In order to assert the improvement obtained by combining two cutting-edge approaches to optimization an extensive set of experiments are carried out. These experiments also test the scalability of MARTEDA as the number of objective functions increases.
@incollection{marti-2011:marteda-lion, address = {Berlin/Heidelberg}, affiliation = {Group of Applied Artificial Intelligence, Universidad Carlos III de Madrid, Av. de la Universidad Carlos III, 22. Colmenarejo, Madrid, 28270 Spain}, author = {Mart\'{i}, Luis and Garc\'{i}a, Jes\'{u}s and Berlanga, Antonio and Molina L\'{o}pez, Jos\'{e} Manuel}, booktitle = {Learning and Intelligent Optimization}, doi = {10.1007/978-3-642-25566-3_36}, editor = {Coello Coello, Carlos A.}, isbn = {978-3-642-25565-6}, keywords = {Computer Science}, pages = {458--472}, publisher = {Springer}, series = {Lecture Notes in Computer Science}, title = {Multi-Objective Optimization with an Adaptive Resonance Theory-Based Estimation of Distribution Algorithm: {A} Comparative Study}, volume = {6683}, year = {2011}, bdsk-url-1 = {https://doi.org/10.1007/978-3-642-25566-3_36} } -
Martí, L., García, J., Berlanga, A., Coello Coello, C. A., & Molina López, J. M. (2011). MB-GNG: Addressing Drawbacks in Multi-Objective Optimization Estimation of Distribution Algorithms. Operations Research Letters, 39(2), 150–154. https://doi.org/10.1016/j.orl.2011.01.002
We examine the model-building issue related to multi-objective estimation of distribution algorithms (MOEDAs) and show that some of their, as yet overlooked, characteristics render most current MOEDAs unviable when addressing optimization problems with many objectives. We propose a novel model-building growing neural gas (MB-GNG) network that is specially devised for properly dealing with that issue and therefore yields a better performance. Experiments are conducted in order to show from an empirical point of view the advantages of the new algorithm.
@article{marti-2011:mb-gng-orl, author = {Mart\'{i}, Luis and Garc\'{i}a, Jes\'{u}s and Berlanga, Antonio and Coello Coello, Carlos~A. and Molina L\'{o}pez, Jos\'{e} Manuel}, doi = {10.1016/j.orl.2011.01.002}, issn = {0167-6377}, journal = {Operations Research Letters}, number = {2}, pages = {150--154}, title = {{MB-GNG}: {A}ddressing Drawbacks in Multi-Objective Optimization Estimation of Distribution Algorithms}, volume = {39}, year = {2011}, bdsk-url-1 = {https://doi.org/10.1016/j.orl.2011.01.002} } -
Wagner, T., Trautmann, H., & Martí, L. (2011). A Taxonomy of Online Stopping Criteria for Multi-Objective Evolutionary Algorithms. In R. H. C. Takahashi, K. Deb, E. F. Wanner, & S. Greco (Eds.), 6th International Conference on Evolutionary Multi-Criterion Optimization (EMO 2011) (Vol. 6576, pp. 16–30). Springer. https://doi.org/10.1007/978-3-642-19893-9_2
The use of multi-objective evolutionary algorithms for solving black-box problems with multiple conflicting objectives has become an important research area. However, when no gradient information is available, the examination of formal convergence or optimality criteria is often impossible. Thus, sophisticated heuristic online stopping criteria (OSC) have recently become subject of intensive research. In order to establish formal guidelines for a systematic research, we present a taxonomy of OSC in this paper. We integrate the known approaches within the taxonomy and discuss them by extracting their building blocks. The formal structure of the taxonomy is used as a basis for the implementation of a comprehensive MATLAB toolbox. Both contributions, the formal taxonomy and the MATLAB implementation, provide a framework for the analysis and evaluation of existing and new OSC approaches.
@inproceedings{wagner-2011:tax, address = {Berlin/Heidelberg}, author = {Wagner, Tobias and Trautmann, Heike and Mart\'{i}, Luis}, booktitle = {6th International Conference on Evolutionary Multi-Criterion Optimization (EMO 2011)}, doi = {10.1007/978-3-642-19893-9_2}, editor = {Takahashi, Ricardo H. C. and Deb, Kalyanmoy and Wanner, Elizabeth F. and Greco, Salvatore}, isbn = {978-3-642-19892-2}, location = {Ouro Preto (MG) Brazil}, pages = {16--30}, publisher = {Springer}, title = {A Taxonomy of Online Stopping Criteria for Multi-Objective Evolutionary Algorithms}, volume = {6576}, year = {2011}, bdsk-url-1 = {https://doi.org/10.1007/978-3-642-19893-9_2} } -
Guerrero, J. L., Martí, L., García, J., Berlanga, A., & Molina López, J. M. (2010). Introducing a Robust and Efficient Stopping Criterion for MOEAs. 2010 IEEE Conference on Evolutionary Computation (CEC), Part of 2010 IEEE World Congress on Computational Intelligence (WCCI 2010). https://doi.org/10.1109/CEC.2010.5586265
Soft computing methods, and Multi-Objective Evolutionary Algorithms (MOEAs) in particular, lack a general convergence criterion which prevents these algorithms from detecting the generation where further evolution will provide little improvements (or none at all) over the current solution, making them waste computational resources. This paper presents the Least Squares Stopping Criterion (LSSC), an easily configurable and implementable, robust and efficient stopping criterion, based on simple statistical parameters and residue analysis, which tries to introduce as few setup parameters as possible, being them always related to the MOEAs research field rather than the techniques applied by the criterion.
@inproceedings{guerrero-2010:lssc, address = {Piscataway, New Jersey}, author = {Guerrero, Jos\'{e} Luis and Mart\'{i}, Luis and Garc\'{i}a, Jes\'{u}s and Berlanga, Antonio and Molina L\'{o}pez, Jos\'{e} Manuel}, booktitle = {2010 IEEE Conference on Evolutionary Computation (CEC), part of 2010 IEEE World Congress on Computational Intelligence (WCCI 2010)}, doi = {10.1109/CEC.2010.5586265}, file = {:papers:guerrero-cec-2010.pdf}, location = {Barcelona, Spain}, publisher = {IEEE Press}, title = {Introducing a Robust and Efficient Stopping Criterion for {MOEAs}}, year = {2010}, bdsk-url-1 = {https://doi.org/10.1109/CEC.2010.5586265} } -
Martí, L., García, J., Berlanga, A., & Molina López, J. M. (2010). Advancing Model-Building for Many-Objective Optimization Estimation of Distribution Algorithms. In C. D. Chio, S. Cagnoni, C. Cotta, M. Ebner, A. Ekárt, A. I. Esparcia-Alcazar, C.-K. Goh, J. J. Merelo, F. Neri, M. Preuß, J. Togelius, & G. N. Yannakakis (Eds.), Applications of Evolutionary Computation (Vol. 6024, pp. 512–521). Springer. https://doi.org/10.1007/978-3-642-12239-2_53
In order to achieve a substantial improvement of MOEDAs regarding MOEAs it is necessary to adapt their model-building algorithms. Most current model-building schemes used so far off-the-shelf machine learning methods. These methods are mostly error-based learning algorithms. However, the model-building problem has specific requirements that those methods do not meet and even avoid. In this work we dissect this issue and propose a set of algorithms that can be used to bridge the gap of MOEDA application. A set of experiments are carried out in order to sustain our assertions.
@inproceedings{marti-2010:advancing-model-building, author = {Mart\'{i}, Luis and Garc\'{i}a, Jes\'{u}s and Berlanga, Antonio and Molina L\'{o}pez, Jos\'{e} Manuel}, booktitle = {Applications of Evolutionary Computation}, doi = {10.1007/978-3-642-12239-2_53}, editor = {Chio, Cecilia Di and Cagnoni, Stefano and Cotta, Carlos and Ebner, Marc and Ek\'{a}rt, Anik\'{o} and Esparcia-Alcazar, Anna I. and Goh, Chi-Keong and Merelo, Juan J. and Neri, Ferrante and Preu{\ss}, Mike and Togelius, Julian and Yannakakis, Georgios N.}, isbn = {978-3-642-12238-5}, location = {Heidelberg/Berlin}, pages = {512--521}, publisher = {Springer}, series = {Lecture Notes in Computer Science}, title = {Advancing Model-Building for Many-Objective Optimization Estimation of Distribution Algorithms}, url = {http://www.springerlink.com/content/cg4nu92878524r23/}, volume = {6024}, year = {2010}, bdsk-url-1 = {http://www.springerlink.com/content/cg4nu92878524r23/}, bdsk-url-2 = {https://doi.org/10.1007/978-3-642-12239-2_53} } -
Martí, L., García, J., Berlanga, A., & Molina López, J. M. (2010). Moving away from error-based learning in multi-objective estimation of distribution algorithms. In J. Branke, E. Alba, D. Arnold, J. Bongard, A. Brabazon, M. V. Butz, J. Clune, M. Cohen, K. Deb, A. Engelbrecht, N. Krasnogor, J. F. Miller, M. O’Neill, K. Sastry, D. Thierens, L. Vanneschi, J. van Hemert, & C. Witt (Eds.), GECCO’10: Proceedings of the 12th Annual Conference on Genetic and Evolutionary Computation (pp. 545–546). ACM Press. https://doi.org/10.1145/1830483.1830585
In this work we analyze the model-building issue and the requirements it imposes on the learning paradigm being used. We argue that error-based learning, the class of learning most commonly used in MOEDAs, is responsible for current MOEDA underachievement. We present ART as a viable alternative and present a novel algorithm called multi-objective ART-based EDA (MARTEDA) that uses a Gaussian ART neural network for model-building and an hypervolume based selector as described for the HypE algorithm. We experimentally show that thanks to MARTEDA’s novel model-building approach and an indicator-based population ranking the algorithm it is able to outperform similar MOEDAs and MOEAs.
@inproceedings{marti-2010:marteda, address = {New York, NY, USA}, author = {Mart\'{i}, Luis and Garc\'{i}a, Jes\'{u}s and Berlanga, Antonio and Molina L\'{o}pez, Jos\'{e} Manuel}, booktitle = {GECCO'10: Proceedings of the 12th Annual Conference on Genetic and Evolutionary Computation}, doi = {10.1145/1830483.1830585}, editor = {Branke, J. and Alba, E. and Arnold, D. and Bongard, J. and Brabazon, A. and Butz, M.~V. and Clune, J. and Cohen, M. and Deb, K. and Engelbrecht, A. and Krasnogor, N. and Miller, J.F. and O'Neill, M. and Sastry, K. and Thierens, D. and Vanneschi, L. and van Hemert, J. and Witt, C.}, isbn = {978-1-4503-0072-8}, location = {Portland, Oregon, USA}, pages = {545--546}, publisher = {ACM Press}, title = {Moving away from error-based learning in multi-objective estimation of distribution algorithms}, url = {http://doi.acm.org/10.1145/1830483.1830585}, year = {2010}, bdsk-url-1 = {http://doi.acm.org/10.1145/1830483.1830585}, bdsk-url-2 = {https://doi.org/10.1145/1830483.1830585} } -
Martí, L., García, J., Berlanga, A., Coello Coello, C. A., & Molina López, J. M. (2010). On Current Model-Building Methods for Multi-Objective Estimation of Distribution Algorithms: Shortcommings and Directions for Improvement (No. GIAA2010E001; Issue GIAA2010E001). Grupo de Inteligencia Artificial Aplicada, Universidad Carlos III de Madrid. http://www.giaa.inf.uc3m.es/miembros/lmarti/model-building
There are some issues with multi-objective estimation of distribution algorithms (MOEDAs) that have been undermining their performance when dealing with problems with many objectives. In this paper we examine the model-building issue related to estimation of distribution algorithms (EDAs) and show that some of their, as yet overlooked, characteristics render most current MOEDAs unviable in the presence of many objectives. First, we present model-building as a problem with particular requirements and explain why some current approaches cannot properly deal with some of these conditions. Then, we discuss the strategies proposed for adapting EDAs to this problem. To validate our working hypothesis, we carry out an experimental study comparing different model-building algorithms. In the final part of the paper, we provide an in-depth discussion on viable alternatives to overcome the limitations of current MOEDAs in many-objective optimization.
@techreport{marti-2010:model-building-tr, address = {Colmenarejo, Spain}, author = {Mart\'{i}, Luis and Garc\'{i}a, Jes\'{u}s and Berlanga, Antonio and Coello Coello, Carlos~A. and Molina L\'{o}pez, Jos\'{e} Manuel}, institution = {Grupo de Inteligencia Artificial Aplicada, Universidad Carlos III de Madrid}, number = {GIAA2010E001}, title = {On Current Model-Building Methods for Multi-Objective Estimation of Distribution Algorithms: Shortcommings and Directions for Improvement}, url = {http://www.giaa.inf.uc3m.es/miembros/lmarti/model-building}, year = {2010}, bdsk-url-1 = {http://www.giaa.inf.uc3m.es/miembros/lmarti/model-building} } -
Martí, L., García, J., Berlanga, A., & Molina López, J. M. (2010). MONEDA: Scalable Multi-Objective Optimization with a Neural Network-based Estimation of Distribution Algorithm (No. GIAA2010E002; Issue GIAA2010E002). Grupo de Inteligencia Artificial Aplicada, Universidad Carlos III de Madrid. http://www.giaa.inf.uc3m.es/miembros/lmarti/moneda
The extension of estimation of distribution algorithms (EDAs) to the multi-objective domain has led to multi-objective optimization EDAs (MOEDAs). Most MOEDAs have limited themselves to porting single-objective EDAs to the multi-objective do- main. Although MOEDAs have proved to be a valid approach, the last point is an obstacle to the achievement of a significant improvement regarding “standard” multi-objective optimization evolutionary algorithms. Adapting the model-building algorithm is one way to achieve a substantial advance. Most model-building schemes used so far by EDAs employ off-the-shelf machine learning methods. However, the model-building problem has particular requirements that those methods do not meet and even evade. The focus of this paper is on the model-building issue and how it has not been prop- erly understood and addressed by most MOEDAs. We delve down into the roots of this matter and hypothesize about its causes. To gain a deeper understanding of the subject we propose a novel algorithm intended to overcome the drawbacks of current MOEDAs. This new algorithm is the multi-objective neural estimation of distribution algorithm (MONEDA). MONEDA uses a modified growing neural gas network for model-building (MB-GNG). MB-GNG is a custom-made clustering algorithm that meets the above demands. Thanks to its custom-made model-building algorithm, the preserva- tion of elite individuals and its individual replacement scheme, MONEDA is capable of scalably solving continuous multi-objective optimization problems. It performs bet- ter than similar algorithms in terms of a set of quality indicators and computational resource requirements.
@techreport{marti-2010:moneda-tr, address = {Colmenarejo, Spain}, author = {Mart\'{i}, Luis and Garc\'{i}a, Jes\'{u}s and Berlanga, Antonio and Molina L\'{o}pez, Jos\'{e} Manuel}, institution = {Grupo de Inteligencia Artificial Aplicada, Universidad Carlos III de Madrid}, number = {GIAA2010E002}, title = {{MONEDA}: {S}calable Multi-Objective Optimization with a Neural Network-based Estimation of Distribution Algorithm}, url = {http://www.giaa.inf.uc3m.es/miembros/lmarti/moneda}, year = {2010}, bdsk-url-1 = {http://www.giaa.inf.uc3m.es/miembros/lmarti/moneda} } -
Martí, L., García, J., Berlanga, A., & Molina López, J. M. (2010). A Progress Indicator for Detecting Success and Failure in Evolutionary Multi–Objective Optimization. 2010 IEEE Conference on Evolutionary Computation (CEC), Part of 2010 IEEE World Congress on Computational Intelligence (WCCI 2010). https://doi.org/10.1109/CEC.2010.5586352
In this work we present a novel progress indicator, called fitness homogeneity indicator (FHI). This indicator improves the other previously discussed indicators as it takes into account all possible processes taking place in the population while not requiring an intensive computation as it relies on the fitness values calculated for the individuals. It is also capable of equally detecting success and failure scenarios, hopefully making an early detection of the second case.
@inproceedings{marti-2010:stability, address = {Piscataway, New Jersey}, author = {Mart\'{i}, Luis and Garc\'{i}a, Jes\'{u}s and Berlanga, Antonio and Molina L\'{o}pez, Jos\'{e} Manuel}, booktitle = {2010 IEEE Conference on Evolutionary Computation (CEC), part of 2010 IEEE World Congress on Computational Intelligence (WCCI 2010)}, doi = {10.1109/CEC.2010.5586352}, file = {:papers:marti-cec-2010.pdf}, location = {Barcelona, Spain}, publisher = {IEEE Press}, title = {A Progress Indicator for Detecting Success and Failure in Evolutionary Multi--Objective Optimization}, year = {2010}, bdsk-url-1 = {https://doi.org/10.1109/CEC.2010.5586352} } -
Guerrero, J. L., García, J., Martí, L., Molina López, J. M., & Berlanga, A. (2009). A Stopping Criterion Based on Kalman Estimation Techniques with Several Progress Indicators. In G. Raidl, E. Alba, J. Bacardit, C. Bates Congdon, H.-G. Beyer, M. Birattari, C. Blum, P. A. N. Bosman, D. Corne, C. Cotta, M. Di Penta, B. Doerr, R. Drechsler, M. Ebner, J. Grahl, T. Jansen, J. Knowles, T. Lenaerts, M. Middendorf, … J. van Hemert (Eds.), GECCO’09: 11th Annual Conference on Genetic and Evolutionary Computation (pp. 587–594). ACM Press. https://doi.org/10.1145/1569901.1569983
The need for a stopping criterion in MOEA’s is a repeatedly mentioned matter in the domain of MOOP’s, even though it is usually left aside as secondary, while stopping criteria are still usually based on an a-priori chosen number of maximum iterations. In this paper we want to present a stopping criterion for MOEA’s based on three different indicators already present in the community. These indicators, some of which were originally designed for solution quality measuring (as a function of the distance to the optimal Pareto front), will be processed so they can be applied as part of a global criterion, based on estimation theory to achieve a cumulative evidence measure to be used in the stopping decision (by means of a Kalman filter). The implications of this cumulative evidence are analyzed, to get a problem and algorithm independent stopping criterion (for each individual indicator). Finally, the stopping criterion is presented from a data fusion perspective, using the different individual indicators’ stopping criteria together, in order to get a final global stopping criterion.
@inproceedings{guerrero-2009:stopping, address = {New York, NY, USA}, author = {Guerrero, Jos\'{e} Luis and Garc\'{i}a, Jes\'{u}s and Mart\'{i}, Luis and Molina L\'{o}pez, Jos\'{e} Manuel and Berlanga, Antonio}, booktitle = {GECCO'09: 11th Annual Conference on Genetic and Evolutionary Computation}, doi = {10.1145/1569901.1569983}, editor = {Raidl, G. and Alba, E. and Bacardit, J. and Bates Congdon, C. and Beyer, H.-G. and Birattari, M. and Blum, C. and Bosman, P.~A.~N. and Corne, D. and Cotta, C. and Di Penta, M. and Doerr, B. and Drechsler, R. and Ebner, M. and Grahl, J. and Jansen, T. and Knowles, J. and Lenaerts, T. and Middendorf, M. and Miller, J.~F. and O'Neill, M. and Poli, R. and Squillero, G. and Stanley, K. and St\"{u}tzle, T. and van Hemert, J.}, isbn = {978-1-60558-325-9}, location = {Montreal, Qu\'{e}bec, Canada}, pages = {587--594}, publisher = {ACM Press}, title = {A Stopping Criterion Based on {K}alman Estimation Techniques with Several Progress Indicators}, url = {http://portal.acm.org/citation.cfm?id=1569983}, year = {2009}, bdsk-url-1 = {http://portal.acm.org/citation.cfm?id=1569983}, bdsk-url-2 = {https://doi.org/10.1145/1569901.1569983} } -
Martí, L., García, J., Berlanga, A., & Molina López, J. M. (2009). On the Computational Properties of the Multi–objective Neural Estimation of Distribution Algorithms. In N. Krasnogor, B. Melián-Batista, J. A. Moreno-Pérez, J. M. Moreno-Vega, & D. Pelta (Eds.), Nature Inspired Cooperative Strategies for Optimization (NICSO 2008) (Vol. 236, pp. 239–251). Springer. https://doi.org/10.1007/978-3-642-03211-0_20
This paper explores the behavior of the multi-objective neural EDA (MONEDA) in terms of its computational requirements it demands and assesses how it scales when dealing with multi-objective optimization problems with relatively large amounts of objectives. In order to properly comprehend these matters other MOEDAs and MOEAs are included in the analysis. The experiments performed tested the ability of each approach to scalably solve many-objective optimization problems. The fundamental result obtained is that MONEDA is not only yields similar or better solutions when compared with other approaches but also does it with at a lower computational cost.
@inproceedings{marti-2008:moneda-comp-cost, address = {Berlin/Heidelberg}, author = {Mart\'{i}, Luis and Garc\'{i}a, Jes\'{u}s and Berlanga, Antonio and Molina L\'{o}pez, Jos\'{e} Manuel}, booktitle = {Nature Inspired Cooperative Strategies for Optimization (NICSO 2008)}, doi = {10.1007/978-3-642-03211-0_20}, editor = {Krasnogor, N. and Meli\'{a}n-Batista, B. and Moreno-P\'{e}rez, J.~A. and Moreno-Vega, J.~M. and Pelta, D.}, isbn = {978-3-642-03210-3}, pages = {239--251}, publisher = {Springer}, series = {Studies in Computational Intelligence}, title = {On the Computational Properties of the Multi--objective Neural Estimation of Distribution Algorithms}, volume = {236}, year = {2009}, bdsk-url-1 = {https://doi.org/10.1007/978-3-642-03211-0_20} } -
Martí, L., García, J., Berlanga, A., & Molina López, J. M. (2009). On the influence of outliers on the loss of population diversity of multi–objective estimation of distribution algorithms. In J. M. Sautto Vallejo, R. Peña Galeana, P. Valdivia Noyola, & N. I. Peña Galeana (Eds.), Taller Latino Iberoamericano de Investigación de Operaciones (TLAIO). Universidad Autónoma de Guerrero.
Multi-objective estimation of distribution algorithms (MOEDAs) have been put forward as a viable alternative to evolutionary algorithms when dealing with complex highly dimensional multi-objective optimization problems. However, these class of algorithms have been shown to exhibit some drawbacks that hinder the achievement of relevant results. One of the most important drawbacks is the gradual loosing of population diversity. This process leads to the deterioration of search capacity of the optimizer and the poor representation of the search space. This loss of diversity could be traced back to the incorrect selection of the algorithms used for building the population model used in current EDAs. These algorithms are off-the-shelf machine learning that are meant to disregard data outliers. The repetitive application of an algorithm that disregards outliers tend to generate more individuals in those zones of the search space that are more densely represented. In this work we investigate this issue comparing different approaches and proposing some viable alternatives. The theoretical discussion is complemented by an extensive experimentation.
@inproceedings{marti-2009:acapulco, address = {Acapulco (Gro), Mexico}, author = {Mart\'{i}, Luis and Garc\'{i}a, Jes\'{u}s and Berlanga, Antonio and Molina L\'{o}pez, Jos\'{e} Manuel}, booktitle = {Taller Latino Iberoamericano de Investigaci\'{o}n de Operaciones (TLAIO)}, editor = {Sautto Vallejo, J. Maclovio and Pe\~na Galeana, Ricardo and Valdivia Noyola, Petra and Pe\~na Galeana, Norma I.}, isbn = {978-607-7760-20-7}, location = {Acapulco (Gro), Mexico}, publisher = {Universidad Aut\'{o}noma de Guerrero}, title = {On the influence of outliers on the loss of population diversity of multi--objective estimation of distribution algorithms}, year = {2009} } -
Martí, L., García, J., Berlanga, A., & Molina López, J. M. (2009). On the Model-Building Issue of Multi-Objective Estimation of Distribution Algorithms. In E. Corchado, X. Wu, E. Oja, álvaro Herrero, & B. Baruque (Eds.), 4th International Conference on Hybrid Artificial Intelligence (HAIS’09) (Vol. 5572, pp. 293–300). Springer. https://doi.org/10.1007/978-3-642-02319-4_35
It has been claimed that perhaps a paradigm shift is necessary in order to be able to deal with this scalability issue of multi-objective optimization evolutionary algorithms. Estimation of distribution algorithms are viable candidates for such task because of their adaptation and learning abilities and simplified algorithmics. Nevertheless, the extension of EDAs to the multi-objective domain have not provided a significant improvement over MOEAs. In this paper we analyze the possible causes of this underachievement and propose a set of measures that should be taken in order to overcome the current situation.
@inproceedings{marti-2009:eda-directions, address = {Berlin/Heidelberg}, author = {Mart\'{i}, Luis and Garc\'{i}a, Jes\'{u}s and Berlanga, Antonio and Molina L\'{o}pez, Jos\'{e} Manuel}, booktitle = {4th International Conference on Hybrid Artificial Intelligence (HAIS'09)}, doi = {10.1007/978-3-642-02319-4_35}, editor = {Corchado, Emilio and Wu, Xindong and Oja, Erkki and Herrero, {\'{a}}lvaro and Baruque, Bruno}, location = {Salamanca, Spain}, pages = {293--300}, publisher = {Springer}, series = {Lecture Notes in Artificial Intelligence}, title = {On the Model-Building Issue of Multi-Objective Estimation of Distribution Algorithms}, volume = {5572}, year = {2009}, bdsk-url-1 = {https://doi.org/10.1007/978-3-642-02319-4_35} } -
Martí, L., García, J., Berlanga, A., & Molina López, J. M. (2009). Solving Complex High-Dimensional Problems with the Multi-Objective Neural Estimation of Distribution Algorithm. In G. Raidl, E. Alba, J. Bacardit, C. Bates Congdon, H.-G. Beyer, M. Birattari, C. Blum, P. A. N. Bosman, D. Corne, C. Cotta, M. Di Penta, B. Doerr, R. Drechsler, M. Ebner, J. Grahl, T. Jansen, J. Knowles, T. Lenaerts, M. Middendorf, … J. van Hemert (Eds.), GECCO’09: 11th Annual Conference on Genetic and Evolutionary Computation (pp. 619–626). ACM Press. https://doi.org/10.1145/1569901.1569987
The multi-objective optimization neural estimation of distribution algorithm (MONEDA) was devised with the purpose of dealing with the model-building issues of MOEDAs and, therefore address their scalability. In this paper we put forward a comprehensive set of experiments that intends to compare MONEDA with similar approaches when solving complex community accepted MOPs. In particular, we deal with the Walking Fish Group scalable test problem set (WFG). These tests aim to establish the optimizing capacity of MONEDA and the consistency as an optimization method. The fundamental conclusion of these assessment is that we provide strong evidences of the viability of MONEDA for handling hard and complex high-dimensional problems and its superior performance when compared to similar approaches. In spite of the fact that obviously further studies are necessary, these extensive experiments have provided solid ground for the use of MONEDA in more ambitious real-world applications.
@inproceedings{marti-2009:moneda-wfg, address = {New York, NY, USA}, author = {Mart\'{i}, Luis and Garc\'{i}a, Jes\'{u}s and Berlanga, Antonio and Molina L\'{o}pez, Jos\'{e} Manuel}, booktitle = {GECCO'09: 11th Annual Conference on Genetic and Evolutionary Computation}, doi = {10.1145/1569901.1569987}, editor = {Raidl, G. and Alba, E. and Bacardit, J. and Bates Congdon, C. and Beyer, H.-G. and Birattari, M. and Blum, C. and Bosman, P.~A.~N. and Corne, D. and Cotta, C. and Di Penta, M. and Doerr, B. and Drechsler, R. and Ebner, M. and Grahl, J. and Jansen, T. and Knowles, J. and Lenaerts, T. and Middendorf, M. and Miller, J. ~F. and O'Neill, M. and Poli, R. and Squillero, G. and Stanley, K. and St\"utzle, T. and van Hemert, J.}, file = {:papers:marti-gecco-2009.pdf}, isbn = {978-1-60558-325-9}, location = {Montreal, Qu\'{e}bec, Canada}, pages = {619--626}, publisher = {ACM Press}, title = {Solving Complex High-Dimensional Problems with the Multi-Objective Neural Estimation of Distribution Algorithm}, url = {http://portal.acm.org/citation.cfm?id=1569901.1569987}, year = {2009}, bdsk-url-1 = {http://portal.acm.org/citation.cfm?id=1569901.1569987}, bdsk-url-2 = {https://doi.org/10.1145/1569901.1569987} } -
Martí, L., García, J., Berlanga, A., & Molina López, J. M. (2009). An Approach to Stopping Criteria for Multi-Objective Optimization Evolutionary Algorithms: The MGBM Criterion. 2009 IEEE Conference on Evolutionary Computation (CEC 2009), 1263–1270. https://doi.org/10.1109/CEC.2009.4983090
In this work we put forward a comprehensive study on the design of global stopping criteria for multi-objective optimization. We describe a novel stopping criterion, denominated MGBM criterion that combines the mutual domination rate (MDR) improvement indicator with a simplified Kalman filter that is used for evidence gathering process. The MDR indicator, which is introduced along, is a special purpose solution meant for the stopping task. It is capable of gauging the progress of the optimization with a low computational cost and therefore suitable for solving complex or many-objective problems. The viability of the proposal is established by comparing it with some other possible alternatives. It should be noted that, although the criteria discussed here are meant for MOPs and MOEAs, they could be easily adapted to other softcomputing or numerical methods by substituting the local improvement metric with a suitable one.
@inproceedings{marti-2009:stopping-cec, address = {Piscataway, New Jersey}, author = {Mart\'{i}, Luis and Garc\'{i}a, Jes\'{u}s and Berlanga, Antonio and Molina L\'{o}pez, Jos\'{e} Manuel}, booktitle = {2009 IEEE Conference on Evolutionary Computation (CEC 2009)}, doi = {10.1109/CEC.2009.4983090}, isbn = {978-1-4244-2959-2}, location = {Trondheim, Norway}, pages = {1263--1270}, publisher = {IEEE Press}, title = {An Approach to Stopping Criteria for Multi-Objective Optimization Evolutionary Algorithms: {T}he {MGBM} Criterion}, year = {2009}, bdsk-url-1 = {https://doi.org/10.1109/CEC.2009.4983090} } -
Fonseca, C., Gandibleux, X., Korhonen, P., Martí, L., Naujoks, B., Thiele, L., Wallenius, J., & Zitzler, E. (2009). Working Group on EMO for Interactive Multiobjective Optimization (1st Round). In K. Deb, S. Greco, K. Miettinen, & E. Zitzler (Eds.), Hybrid and Robust Approaches to Multiobjective Optimization (Issue 09041). Schloss Dagstuhl — Leibniz–Zentrum fuer Informatik. http://drops.dagstuhl.de/opus/volltexte/2009/2004
This group explored the use of EMO in an interactive manner to solve multiobjective optimization problems.
@inproceedings{marti-dag-2009, address = {Dagstuhl, Germany}, author = {Fonseca, Carlos and Gandibleux, Xavier and Korhonen, Pekka and Mart\'{i}, Luis and Naujoks, Boris and Thiele, Lothar and Wallenius, Jyrki and Zitzler, Eckart}, booktitle = {Hybrid and Robust Approaches to Multiobjective Optimization}, editor = {Deb, Kalyanmoy and Greco, Salvatore and Miettinen, Kaisa and Zitzler, Eckart}, issn = {1862-4405}, number = {09041}, publisher = {Schloss Dagstuhl --- Leibniz--Zentrum fuer Informatik}, series = {Dagstuhl Seminar Proceedings}, title = {{W}orking Group on {EMO} for Interactive Multiobjective Optimization (1st Round)}, url = {http://drops.dagstuhl.de/opus/volltexte/2009/2004}, year = {2009}, bdsk-url-1 = {http://drops.dagstuhl.de/opus/volltexte/2009/2004} } -
Martí, L., García, J., Berlanga, A., & Molina López, J. M. (2008). Scalable Continuous Multiobjective Optimization with a Neural Network–Based Estimation of Distribution Algorithm. In M. Giacobini, A. Brabazon, S. Cagnoni, G. A. Di Caro, R. Drechsler, A. Ekárt, A. I. Esparcia-Alcázar, M. Farooq, A. Fink, J. McCormack, M. O’Neill, J. Romero, F. Rothlauf, G. Squillero, A. Ş. Uyar, & S. Yang (Eds.), Applications of Evolutionary Computing (Vol. 4974, pp. 535–544). Springer. https://doi.org/10.1007/978-3-540-78761-7_59
To achieve a substantial improvement of MOEDAs regarding MOEAs it is necessary to adapt their model building algorithm to suit this particular task. Most current model building schemes used so far off-the-shelf machine learning methods. However, the model building problem has specific requirements that those methods do not meet and even avoid. In this we work propose a novel approach to model building in MOEDAs using an algorithm custom-made for the task. We base our proposal on the growing neural gas (GNG) network. The resulting model-building GNG (MB-GNG) is capable of yielding good results when confronted to high-dimensional problems.
@inproceedings{marti-2008:evonum, address = {Heidelberg}, author = {Mart\'{i}, Luis and Garc\'{i}a, Jes\'{u}s and Berlanga, Antonio and Molina L\'{o}pez, Jos\'{e} Manuel}, booktitle = {Applications of Evolutionary Computing}, doi = {10.1007/978-3-540-78761-7_59}, editor = {Giacobini, Mario and Brabazon, Anthony and Cagnoni, Stefano and Di Caro, Gianni~A. and Drechsler, Rolf and Ek\'{a}rt, Anik\'{o} and Esparcia-Alc\'{a}zar, Anna Isabel and Farooq, Muddassar and Fink, Andreas and McCormack, Jon and O'Neill, Michael and Romero, Juan and Rothlauf, Franz and Squillero, Giovanni and Uyar, A. \c{S}ima and Yang, Shengxiang}, isbn = {978-3-540-78760-0}, pages = {535--544}, publisher = {Springer}, series = {Lecture Notes in Computer Science}, title = {Scalable Continuous Multiobjective Optimization with a Neural Network--Based Estimation of Distribution Algorithm}, volume = {4974}, year = {2008}, bdsk-url-1 = {https://doi.org/10.1007/978-3-540-78761-7_59} } -
Martí, L., García, J., Berlanga, A., & Molina López, J. M. (2008). Model-Building Algorithms for Multiobjective EDAs: Directions for Improvement. 2008 IEEE Conference on Evolutionary Computation (CEC), Part of 2008 IEEE World Congress on Computational Intelligence (WCCI 2008), 2848–2855. https://doi.org/10.1109/CEC.2008.4631179
In order to comprehend the advantages and short-comings of each model-building algorithm they should be tested under similar conditions and isolated from the MOEDA it takes part of. In this work we will assess some of the main machine learning algorithms used or suitable for model-building in a controlled environment and under equal conditions. They are analyzed in terms of solution accuracy and computational complexity. To the best of our knowledge a study like this has not been put forward before and it is essential for the understanding of the nature of the model-building problem of MOEDAs and how they should be improved to achieve a quantum leap in their problem solving capacity.
@inproceedings{marti-2008:model-comp, address = {Piscataway, New Jersey}, author = {Mart\'{i}, Luis and Garc\'{i}a, Jes\'{u}s and Berlanga, Antonio and Molina L\'{o}pez, Jos\'{e} Manuel}, booktitle = {2008 IEEE Conference on Evolutionary Computation (CEC), part of 2008 IEEE World Congress on Computational Intelligence (WCCI 2008)}, doi = {10.1109/CEC.2008.4631179}, isbn = {978-1-4244-1823-7}, location = {Hong Kong, China}, pages = {2848--2855}, publisher = {IEEE Press}, title = {Model-Building Algorithms for Multiobjective {EDA}s: {D}irections for Improvement}, url = {http://ieeexplore.ieee.org/iel5/4625778/4630767/04631179.pdf?tp=&arnumber=4631179&isnumber=4630767}, year = {2008}, bdsk-url-1 = {http://ieeexplore.ieee.org/iel5/4625778/4630767/04631179.pdf?tp=&arnumber=4631179&isnumber=4630767}, bdsk-url-2 = {https://doi.org/10.1109/CEC.2008.4631179} } -
Martí, L., García, J., Berlanga, A., & Molina López, J. M. (2008). Introducing MONEDA: Scalable Multiobjective Optimization with a Neural Estimation of Distribution Algorithm. In M. Keizer, G. Antoniol, C. Congdon, K. Deb, B. Doerr, N. Hansen, J. Holmes, G. Hornby, D. Howard, J. Kennedy, S. Kumar, F. Lobo, J. Miller, J. Moore, F. Neumann, M. Pelikan, J. Pollack, K. Sastry, K. Stanley, … I. Wegener (Eds.), GECCO’08: 10th Annual Conference on Genetic and Evolutionary Computation (pp. 689–696). ACM Press. https://doi.org/10.1145/1389095.1389230
In this paper we explore the model-building issue of multiobjective optimization estimation of distribution algorithms. We argue that model-building has some characteristics that differentiate it from other machine learning tasks. A novel algorithm called multiobjective neural estimation of distribution algorithm (MONEDA) is proposed to meet those characteristics. This algorithm uses a custom version of the growing neural gas (GNG) network specially meant for the model-building task. As part of this work, MONEDA is assessed with regard to other classical and state-of-the-art evolutionary multiobjective optimizers when solving some community accepted test problems.
@inproceedings{marti-2008:moneda, address = {New York, NY, USA}, author = {Mart\'{i}, Luis and Garc\'{i}a, Jes\'{u}s and Berlanga, Antonio and Molina L\'{o}pez, Jos\'{e} Manuel}, booktitle = {GECCO'08: 10th Annual Conference on Genetic and Evolutionary Computation}, doi = {10.1145/1389095.1389230}, editor = {Keizer, Marten and Antoniol, Giulio and Congdon, Clare and Deb, Kalyanmoy and Doerr, Benjamin and Hansen, Nikolaus and Holmes, John and Hornby, Gergory and Howard, Daniel and Kennedy, John and Kumar, Sanjeev and Lobo, Ferdinando and Miller, Julian and Moore, Jason and Neumann, Frank and Pelikan, Martin and Pollack, Jordan and Sastry, Kumara and Stanley, Ken and Stoica, Adrian and Talbi, El Ghazli and Wegener, Ingo}, isbn = {978-1-60558-131-6}, location = {Atlanta (GA), USA}, note = {EMO Track ``Best Paper'' Nominee}, pages = {689--696}, publisher = {ACM Press}, title = {Introducing {MONEDA}: {S}calable Multiobjective Optimization with a Neural Estimation of Distribution Algorithm}, year = {2008}, bdsk-url-1 = {https://doi.org/10.1145/1389095.1389230} } -
Martí, L., García, J., Berlanga, A., & Molina López, J. M. (2007). A Cumulative Evidential Stopping Criterion for Multiobjective Optimization Evolutionary Algorithms. In D. Thierens, K. Deb, M. Pelikan, H.-G. Beyer, B. Doerr, R. Poli, & M. Bittari (Eds.), GECCO’07: Proceedings of the 2007 GECCO Conference Companion on Genetic and Evolutionary Computation (p. 911). ACM Press. https://doi.org/10.1145/1276958.1277141
In this work we present a novel and efficient algorithm independent stopping criterion, called the MGBM criterion,suitable for Multi-objective Optimization Evolutionary Algorithms (MOEAs).The criterion, after each iteration of the optimization algorithm, gathers evidence of the improvement of the solutions obtained so far. A global (execution wise) evidence accumulation process inspired by recursive Bayesian estimation decides when the optimization should be stopped. Evidence is collected using a novel relative improvement measure constructed on top of the Pareto dominance relations. The evidence gathered after each iteration is accumulated and updated following a rule based on a simplified version of a discrete Kalman filter.Our criterion is particularly useful in complex and/or high-dimensional problems where the traditional procedure of stopping after a predefined amount of iterations cannot be used and the waste of computational resources can induce to a detriment of the quality of the results.Although the criterion discussed here is meant for MOEAs,it can be easily adapted to other soft computing or numerical methods by substituting the local improvement metric with a suitable one.
@inproceedings{marti-2007:gecco-stopping, address = {New York}, author = {Mart\'{i}, Luis and Garc\'{i}a, Jes\'{u}s and Berlanga, Antonio and Molina L\'{o}pez, Jos\'{e} Manuel}, bdsk-url-2 = {https://doi.org/10.1145/1276958.1277141}, booktitle = {GECCO'07: Proceedings of the 2007 GECCO Conference Companion on Genetic and Evolutionary Computation}, doi = {10.1145/1276958.1277141}, editor = {Thierens, Dirk and Deb, Kalyanmoy and Pelikan, Martin and Beyer, Hans-Georg and Doerr, Benjamin and Poli, Riccardo and Bittari, Mauro}, isbn = {978-1-59593-697-4}, location = {London, UK}, pages = {911}, publisher = {ACM Press}, title = {A Cumulative Evidential Stopping Criterion for Multiobjective Optimization Evolutionary Algorithms}, url = {http://portal.acm.org/citation.cfm?doid=1276958.1277141}, year = {2007}, bdsk-url-1 = {http://portal.acm.org/citation.cfm?doid=1276958.1277141} } -
Martí, L., García, J., Berlanga, A., & Molina López, J. M. (2007). A Cumulative Evidential Stopping Criterion for Multiobjective Optimization Evolutionary Algorithms (extended version). In D. Thierens, K. Deb, M. Pelikan, H.-G. Beyer, B. Doerr, R. Poli, & M. Bittari (Eds.), GECCO’07: Proceedings of the 2007 GECCO Conference Companion on Genetic and Evolutionary Computation (pp. 2835–2842). ACM. https://doi.org/10.1145/1274000.1274053
In this work we present a novel and efficient algorithm independent stopping criterion, called the MGBM criterion, suitable for Multiobjective Optimization Evolutionary Algorithms (MOEAs). The criterion, after each iteration of the optimization algorithm, gathers evidence of the improvement of the solutions obtained so far. A global (execution wise) evidence accumulation process inspired by recursive Bayesian estimation decides when the optimization should be stopped. Evidence is collected using a novel relative improvement measure constructed on top of the Pareto dominance relations. The evidence gathered after each iteration is accumulated and updated following a rule based on a simplified version of a discrete Kalman filter. Our criterion is particularly useful in complex and/or high-dimensional problems where the traditional procedure of stopping after a predefined amount of iterations cannot be used and the waste of computational resources can induce to a detriment of the quality of the results. Although the criterion discussed here is meant for MOEAs, it can be easily adapted to other soft computing or numerical methods by substituting the local improvement metric with a suitable one.
@inproceedings{marti-2007:gecco-stopping-extended, address = {New York, NY, USA}, author = {Mart\'{i}, Luis and Garc\'{i}a, Jes\'{u}s and Berlanga, Antonio and Molina L\'{o}pez, Jos\'{e} Manuel}, booktitle = {GECCO'07: Proceedings of the 2007 GECCO Conference Companion on Genetic and Evolutionary Computation}, doi = {10.1145/1274000.1274053}, editor = {Thierens, Dirk and Deb, Kalyanmoy and Pelikan, Martin and Beyer, Hans-Georg and Doerr, Benjamin and Poli, Riccardo and Bittari, Mauro}, isbn = {978-1-59593-698-1}, location = {London, United Kingdom}, pages = {2835--2842}, publisher = {ACM}, title = {A Cumulative Evidential Stopping Criterion for Multiobjective Optimization Evolutionary Algorithms (extended version)}, year = {2007}, bdsk-url-1 = {https://doi.org/10.1145/1274000.1274053} } -
Martí, L. (2007). A Hybrid Neural System based on Adaptive Resonance Theory and Representational Redescription capable of Variable Binding. In J. Si & R. Sun (Eds.), 2007 International Joint Conference on Neural Networks (IJCNN) (pp. 2448–2453). IEEE Press. https://doi.org/10.1109/IJCNN.2007.4371342
In this work we propose a hybrid neural architecture named VaBiARRT based on the adaptive resonance theory that relies on representational redescription to archive a high degree of generalization and code compactness. Knowledge is autonomously structured as a hierarchical topology of fuzzy input classes and how these classes are related with the outputs. A two-way abstraction/particularization process takes place in order to rewrite the established relations to make them as abstract as possible without loosing accuracy. The internal handling of the representation of knowledge can be interpreted as a standard pattern matching variable binding process. Besides providing a formal description of VaBiARRT we also solve a sample problem related to the representational redescription hypothesis in order to study the knowledge redescription process and the performance of the network.
@inproceedings{marti-2007:vabbiart, author = {Mart\'{i}, Luis}, booktitle = {2007 International Joint Conference on Neural Networks (IJCNN)}, doi = {10.1109/IJCNN.2007.4371342}, editor = {Si, Jennie and Sun, Ron}, issn = {1098-7576}, pages = {2448--2453}, publisher = {IEEE Press}, title = {A Hybrid Neural System based on Adaptive Resonance Theory and Representational Redescription capable of Variable Binding}, url = {http://www.ieeexplore.ieee.org/xpls/abs\%5Fall.jsp?isnumber=4370891&arnumber=4371342&count=569&index=450}, year = {2007}, bdsk-url-1 = {http://www.ieeexplore.ieee.org/xpls/abs%5C%5Fall.jsp?isnumber=4370891&arnumber=4371342&count=569&index=450}, bdsk-url-2 = {https://doi.org/10.1109/IJCNN.2007.4371342} } -
Molina López, J. M., Chamorro, F., Ledezma, A., Carbó, J., Martí, L., Pérez, Ó., & García, J. (2006). Fundamentos de Programación: Grado Superior (Fundamentals of Programming). McGraw-Hill. http://www.mcgraw-hill.es/html/8448148681.html
With a contemporary, engaging, and practical approach, this editorial project develops the contents of the Programming Fundamentals module, included in the Advanced Vocational Training Course in Computer Systems Administration. The main objective of the book is to develop the basic concepts of programming related to algorithms and data structures. The book covers control structures and the various data structures used in software problem-solving. Texts on programming devote a significant portion of their content to explaining theoretical concepts, leaving the student with a small collection of exercises or case studies. In this text, the authors have reversed this trend and proposed practical examples that students can program to gain a practical understanding of the previously explained theoretical content. By acquiring this knowledge, students are satisfactorily prepared to enter the labor market, being able to perform the following functions: choose and define a data structure to solve a problem with structured languages; apply the structured development methodology for algorithm development; and code programs in third-generation structured languages.
@book{molina-et-al-2006:fpgs, address = {Madrid}, author = {Molina L\'{o}pez, Jos\'{e} Manuel and Chamorro, F\'{e}lix and Ledezma, Agapito and Carb\'{o}, Javier and Mart\'{i}, Luis and P\'{e}rez, \'{O}scar and Garc\'{i}a, Jes\'{u}s}, isbn = {8448148681}, publisher = {McGraw-Hill}, title = {Fundamentos de Programaci{\'{o}}n: Grado Superior (\emph{Fundamentals of Programming})}, url = {http://www.mcgraw-hill.es/html/8448148681.html}, year = {2006}, bdsk-url-1 = {http://www.mcgraw-hill.es/html/8448148681.html} } -
Molina López, J. M., Chamorro, F., Ledezma, A., Carbó, J., Martí, L., Pérez, Ó., & García, J. (2006). Guía Didáctica: Fundamentos de Programación: Grado Superior (Instructor’s Guide: Fundamentals of Programming). McGraw–Hill. http://www.mcgraw-hill.es/html/8448148681.html
Instructor guide for book “Fundamentos de Programación (Grado Superior).”
@book{molina-et-al-2006:gdfpgs, address = {Madrid}, author = {Molina L\'{o}pez, Jos\'{e} Manuel and Chamorro, F\'{e}lix and Ledezma, Agapito and Carb\'{o}, Javier and Mart\'{i}, Luis and P\'{e}rez, \'{O}scar and Garc\'{i}a, Jes\'{u}s}, isbn = {8448148681}, publisher = {McGraw--Hill}, title = {Gu\'{i}a Did{\'{a}}ctica: {F}undamentos de Programaci{\'{o}}n: Grado Superior (\emph{Instructor's Guide: Fundamentals of Programming})}, url = {http://www.mcgraw-hill.es/html/8448148681.html}, year = {2006}, bdsk-url-1 = {http://www.mcgraw-hill.es/html/8448148681.html} } -
Molina López, J. M., Chamorro, F., Ledezma, A., Carbó, J., Martí, L., Pérez, Ó., & García, J. (2006). Guía Didáctica: Programación en Lenguajes Estructurados: Grado Superior (Instructor’s Guide for Programming in Strucured Languages). McGraw-Hill. http://www.mcgraw-hill.es/html/8448148711.html
Instructor guide for book “Programación en Lenguajes Estructurados (Grado Superior).”
@book{molina-et-al-2006:gdplegs, address = {Madrid}, author = {Molina L\'{o}pez, Jos\'{e} Manuel and Chamorro, F\'{e}lix and Ledezma, Agapito and Carb\'{o}, Javier and Mart\'{i}, Luis and P\'{e}rez, \'{O}scar and Garc\'{i}a, Jes\'{u}s}, isbn = {8448148711}, publisher = {McGraw-Hill}, title = {Gu\'{i}a Did{\'{a}}ctica: {P}rogramaci{\'{o}}n en Lenguajes Estructurados: Grado Superior (\emph{Instructor's Guide for Programming in Strucured Languages})}, url = {http://www.mcgraw-hill.es/html/8448148711.html}, year = {2006}, bdsk-url-1 = {http://www.mcgraw-hill.es/html/8448148711.html} } -
Molina López, J. M., Chamorro, F., Ledezma, A., Carbó, J., Martí, L., Pérez, Ó., & García, J. (2006). Programación en Lenguajes Estructurados: Grado Superior (Programming in Strucured Languages). McGraw-Hill. http://www.mcgraw-hill.es/html/8448148703.html
With a contemporary, engaging, and practical approach, this publishing project develops the contents of the Structured Language Programming module, included in the Advanced Vocational Training Course in Computer Application Development. The main objective of this book is to develop the most important concepts in the program creation process, from the basic structures of imperative programming to the more abstract concepts of object-oriented design. The book covers the control structures and the various data structures used in software problem-solving. The language used to explain all these ideas is, first, C, through which the basic rudiments of imperative programming are learned and where the concept of structured programming is introduced. C++ is then introduced to lead to the JavaTM language. This sequence of languages makes the book especially interesting for first-year students starting from scratch or for those who, despite having knowledge of other programming languages, are about to make the leap to programming in JavaTM. By acquiring this knowledge, students will be well prepared to enter the workforce, enabling them to perform the following functions: develop programs using structured languages that comply with the specifications established in the design; evaluate the functionality of applications by testing the various programming modules; prepare complete documentation for the developed applications; and adapt applications based on new requirements established in the design.
@book{molina-et-al-2006:plegs, address = {Madrid}, author = {Molina L\'{o}pez, Jos\'{e} Manuel and Chamorro, F\'{e}lix and Ledezma, Agapito and Carb\'{o}, Javier and Mart\'{i}, Luis and P\'{e}rez, \'{O}scar and Garc\'{i}a, Jes\'{u}s}, isbn = {8448148703}, publisher = {McGraw-Hill}, title = {Programaci{\'{o}}n en Lenguajes Estructurados: Grado Superior (\emph{Programming in Strucured Languages})}, url = {http://www.mcgraw-hill.es/html/8448148703.html}, year = {2006}, bdsk-url-1 = {http://www.mcgraw-hill.es/html/8448148703.html} } -
Pérez, M. J., García, J., Martí, L., & Molina López, J. M. (2006). Multi-objective Optimization Evolutionary Algorithms in Insurance Linked Derivatives. In J.-P. Rennard (Ed.), Handbook of Research on Nature-inspired Computing for Economics and Management: Vol. II (pp. 885–908). Idea Group. https://doi.org/10.4018/978-1-59140-984-7.ch057
This work addresses a real-world adjustment of economic models where the application of robust and global optimization techniques is required. The problem dealt with is the search for a set of parameters to calculate the reported claim amount. Several functions are proposed to obtain the reported claim amount, and a multi-objective optimization procedure is used to obtain parameters using real data and to decide the best function to approximate the reported claim amount. Using this function, insurance companies negotiate the underlying contract-that is, the catastrophic loss ratio defined from the total reported claim amount. They are associated with catastrophes that occurred during the loss period and declared until the development period expired. The suitability of different techniques coming from evolutionary computation (EC) to solve this problem is explored, contrasting the performance achieved with recent proposals of multi-objective evolutionary algorithms (MOEAs). Results show the advantages of MOEAs in the proposal in terms of effectiveness and completeness in searching for solutions, compared with particular solutions of classical EC approaches (using an aggregation operator) in problems with real data.
@incollection{perez-et-al-2006:moea-in-insurance, address = {London}, author = {P\'{e}rez, Mar\'{i}a Jos\'{e} and Garc\'{i}a, Jes\'{u}s and Mart\'{i}, Luis and Molina L\'{o}pez, Jos\'{e} Manuel}, booktitle = {Handbook of Research on Nature-inspired Computing for Economics and Management}, doi = {10.4018/978-1-59140-984-7.ch057}, editor = {Rennard, J.-P.}, pages = {885--908}, publisher = {Idea Group}, title = {Multi-objective Optimization Evolutionary Algorithms in Insurance Linked Derivatives}, url = {http://www.igi-global.com/bookstore/Chapter.aspx?TitleId=21172}, volume = {II}, year = {2006}, bdsk-url-1 = {http://www.igi-global.com/bookstore/Chapter.aspx?TitleId=21172}, bdsk-url-2 = {https://doi.org/10.4018/978-1-59140-984-7.ch057} } -
Martí, L., Policriti, A., & García, L. (2004). A Hybrid ART Neuro–fuzzy Architecture with Variable Binding. Proceedings of the First Cuban Artificial Intelligence Symposium.
@inproceedings{marti-2004:hybrid-art, address = {La Habana, Cuba}, author = {Mart\'{i}, Luis and Policriti, Alberto and Garc\'{i}a, Luciano}, booktitle = {Proceedings of the First Cuban Artificial Intelligence Symposium}, publisher = {Universidad de las Ciencias Inform\'{a}ticas}, title = {A Hybrid {ART} Neuro--fuzzy Architecture with Variable Binding}, year = {2004} } -
Martí, L., Policriti, A., & García, L. (2003). Redes ART híbridas para la predicción de series de tiempo. In M. García (Ed.), Resúmenes del 8vo Congreso Nacional de Matemáticas y Computación (COMPUMAT’2003). Sociedad Cubana de Matemáticas y Computación.
@inproceedings{marti-2003:art-series, author = {Mart\'{i}, Luis and Policriti, Alberto and Garc\'{i}a, Luciano}, booktitle = {Res\'{u}menes del 8vo Congreso Nacional de Matem\'{a}ticas y Computaci\'{o}n (COMPUMAT'2003)}, editor = {Garc\'{i}a, Mauro}, publisher = {Sociedad Cubana de Matem\'{a}ticas y Computaci\'{o}n}, title = {Redes {ART} h\'{i}bridas para la predicci\'{o}n de series de tiempo}, year = {2003} } -
Martí, L., Policriti, A., & García, L. (2003). Modelos Neurodifusos Híbridos Basados en la Teoría de Resonancia Adaptativa (Neuro-fuzzy Hybrid Models based on Adaptive Resonance Theory). In G. Joya, M. A. Atencia, A. Ochoa, & S. Allende (Eds.), Optimización Inteligente (pp. 363–412). Servicio de Publicaciones de la Universidad de Málaga (SPICUM). http://malaka.spicum.uma.es/libro.php?idLibro=909
This chapter discusses models that combine the characteristics of artificial neural networks, neurofuzzy models, and symbolic artificial intelligence. A commentary on several models is presented, and their results in solving various practical problems are analyzed.
@incollection{marti-2003:malaga-book, address = {M\'{a}laga}, author = {Mart\'{i}, Luis and Policriti, Alberto and Garc\'{i}a, Luciano}, booktitle = {Optimizaci\'{o}n Inteligente}, editor = {Joya, Gonzalo and Atencia, M.~A. and Ochoa, Alberto and Allende, Sira}, isbn = {84-9747-034-6}, pages = {363--412}, publisher = {Servicio de Publicaciones de la Universidad de M\'{a}laga (SPICUM)}, title = {Modelos Neurodifusos H\'{i}bridos Basados en la Teor\'{i}a de Resonancia Adaptativa (Neuro-fuzzy Hybrid Models based on Adaptive Resonance Theory)}, url = {http://malaka.spicum.uma.es/libro.php?idLibro=909}, year = {2003}, bdsk-url-1 = {http://malaka.spicum.uma.es/libro.php?idLibro=909} } -
Martí, L., Policriti, A., & García, L. (2003). Hybrid Adaptive Resonance Theory neural networks for function approximation. In A. Abraham, L. C. Jain, & B. J. van der Zwaag (Eds.), Innovations in Intelligent Systems and Applications: Design, Management and Applications (pp. 51–88). Physica–Verlag (Springer). http://www.springer.com/engineering/book/978-3-540-20265-3
AppART is an adaptive resonance theory low parameterized neural model that incrementally approximates continuous—valued multidimensional functions from noisy data using biologically plausible processes. AppART performs a higher—order Nadaraya—Watson regression and can be interpreted as a fuzzy logic standard additive model. In this chapter we describe AppART dynamics and training. We discuss the approach it makes to hybrid neural systems and deal with its theoretical foundations as a function approximation method. Two modifications to AppART, aimed at improving AppART efficiency, are proposed and tested. We also discuss the combination AppART with growing neural gas networks. Finally, four benchmark problems are solved in order to study AppART from a practical point of view and to compare its results with those obtained from other models.
@incollection{marti-book-chapter, address = {Heidelberg}, author = {Mart\'{i}, Luis and Policriti, Alberto and Garc\'{i}a, Luciano}, booktitle = {Innovations in Intelligent Systems and Applications: Design, Management and Applications}, editor = {Abraham, Ajith and Jain, Lakmi C. and van der Zwaag, Berend Jan}, isbn = {978-3-540-20265-3}, pages = {51--88}, publisher = {Physica--Verlag (Springer)}, series = {Studies in Fuzziness and Soft Computing}, title = {Hybrid Adaptive Resonance Theory neural networks for function approximation}, url = {http://www.springer.com/engineering/book/978-3-540-20265-3}, year = {2003}, bdsk-url-1 = {http://www.springer.com/engineering/book/978-3-540-20265-3} } -
Martí, L., Policriti, A., & García, L. (2002). AppART: An ART Hybrid Stable Learning Neural Network for Universal Function Approximation. In A. Abraham & M. Koeppen (Eds.), Hybrid Information Systems (pp. 93–120). Physica–Verlag. https://doi.org/10.1007/978-3-7908-1782-9_9
This work describes AppART, an ART–based low parameterized neural model that incrementally approximates continuous–valued multidimensional functions from noisy data using biologically plausible processes. AppART performs a higher–order Nadaraya–Watson regression and can be interpreted as a fuzzy system. Some benchmark problems are solved in order to study AppART from an application point of view and to compare its results with the ones obtained from other models.
@inproceedings{appart-his-2002, address = {Heidelberg}, author = {Mart\'{i}, Luis and Policriti, Alberto and Garc\'{i}a, Luciano}, booktitle = {Hybrid Information Systems}, doi = {10.1007/978-3-7908-1782-9_9}, editor = {Abraham, Ajith and Koeppen, Mario}, pages = {93--120}, publisher = {Physica--Verlag}, title = {{AppART}: {A}n {ART} Hybrid Stable Learning Neural Network for Universal Function Approximation}, url = {http://www.springer.com/computer/artificial/book/978-3-7908-1480-4}, year = {2002}, bdsk-url-1 = {http://www.springer.com/computer/artificial/book/978-3-7908-1480-4}, bdsk-url-2 = {https://doi.org/10.1007/978-3-7908-1782-9_9} } -
Martí, L., Policriti, A., García, L., & Lazo, R. (2002). AppART + Growing Neural Gas = High Performance Hybrid Neural Network for Function Approximation. In M. Bhattacharya & A. Abraham (Eds.), Workshop on Intelligent Knowledge Management Techniques (IKOMAT’2002), part of Knowledge-based Intelligent Information Engineering Systems & Allied Technologies (KES’2002) (pp. 1483–1487). IOS Press. http://www.iospress.nl/loadtop/load.php?isbn=9781586032807
We combine two notable streams of neural networks research: Adaptive Resonance Theory (ART) and Growing Neural Gas (GNG) networks. In particular we modify the AppART neural network formulation by introducing GNG based training features. The resulting neural network outperforms its original version as well as other neural models while maintaining the functional approximation properties and hybrid system conception.
@inproceedings{Marti-GasART02, address = {Amsterdam}, author = {Mart\'{i}, Luis and Policriti, Alberto and Garc\'{i}a, Luciano and Lazo, Raynel}, booktitle = {Workshop on Intelligent Knowledge Management Techniques (IKOMAT'2002), part of Knowledge-based Intelligent Information Engineering Systems \& Allied Technologies (KES'2002)}, editor = {Bhattacharya, M. and Abraham, A.}, isbn = {978-1-58603-280-7}, pages = {1483--1487}, publisher = {IOS Press}, title = {{AppART} + Growing Neural Gas = High Performance Hybrid Neural Network for Function Approximation}, url = {http://www.iospress.nl/loadtop/load.php?isbn=9781586032807}, year = {2002}, bdsk-url-1 = {http://www.iospress.nl/loadtop/load.php?isbn=9781586032807} } -
Martí, L., Catasús, M., & García, L. (1999). Artificial Neural Networks in Analytical Chemistry. CIMAF’99: International Conference Science and Technology for Development — Adaptive Systems Symposium. https://scienceon.kisti.re.kr/srch/selectPORSrchArticle.do?cn=NPAP00983174
We present a novel application of artificial neural networks on the context of analytical chemistry applications.
@inproceedings{marti-1999:cimaf, author = {Mart\'{i}, Luis and Catas\'{u}s, Miguel and Garc\'{i}a, Luciano}, booktitle = {{CIMAF'99}: International Conference Science and Technology for Development --- Adaptive Systems Symposium}, title = {Artificial Neural Networks in Analytical Chemistry}, url = {https://scienceon.kisti.re.kr/srch/selectPORSrchArticle.do?cn=NPAP00983174}, year = {1999}, bdsk-url-1 = {https://scienceon.kisti.re.kr/srch/selectPORSrchArticle.do?cn=NPAP00983174} }