Contact

victoruccetina@gmail.com

Experience

September 2002 - Currently (20 years): Professor of Computer Science in the Facultad de Matemáticas at Universidad Autónoma de Yucatán. Mexico.

Sep 2016 - Aug 2017,  Apr 2019 - Sep 2020, Apr 2022 - Sep 2022 (3 years): Visiting Professor of Machine Learning in the Fachbereich Informatik at Universität Hamburg. Germany.

October 2011 - March 2012 (6 months): Postdoctoral Researcher on Machine Learning in the School of Computer Science at The University of Manchester. United Kingdom.

January 2006 - March 2006 (3 months): Visiting Doctoral Student in the Autonomous Learning Laboratory at University of Massachusetts Amherst. United States.

Education

Ph.D. Computer Science (Dr. rer. nat.), Humboldt-Universität zu Berlin, Germany, 2009.

M.Sc. Intelligent Systems, Instituto Tecnológico y de Estudios Superiores de Monterrey, México, 2002.

B.Sc. Software Engineering, Instituto Tecnológico de Mérida, México, 1998.

Supervised theses

PhD Theses 

• Saúl Hazel Martínez Treviño. Aprendizaje automático para elucidación de estructuras moleculares. Cinvestav Mérida. 2022.

MSc Theses

• Dayan Bravo Fraga. Pointing gesture recognition. 2024.

• Mariana Inés Torres Pozos. Predicción de frames en secuencia de imágenes. 2023.

• Laura Álvarez González. Generación de imágenes usando redes adversarias generativas para aumentación de datos de entrenamiento de modelos de re-identificación. 2023.

• Oliver Bartels. Adjusting an intrinsically motivated RL controller for simulated car racing. University of Hamburg. 2021.

• Anton Volkov. On the influence of positionally-aware encoding methods on ranking performance for ambiguous queries. University of Hamburg. 2021.

• Amer Althiab. On stability analysis and performance of neural networks gated recurrent unit (GRU) with or without attention mechanisms. University of Hamburg. 2021.

• Iryna Tyshko. On the performance and stability of training GANs as dynamical systems using MMD metric with an application of the attention mechanisms in image generation. University of Hamburg. 2021.

• Fabian Hausmann. Recurrent neural networks for the prediction of genomic repetitive elements. University of Hamburg. 2019.

• Juan José Negrón Granados. Detección del árbol de dzidzilché usando imágenes aéreas multiespectrales y redes neuronales convolucionales. Universidad Autónoma de Yucatán. 2019.

• Iván de Jesús Martínez Chin. Red neuronal con convolución dilatada para estimación de flujo óptico denso. Universidad Autónoma de Yucatán. 2019.

• Jochen Sieg. Evaluation of benchmark datasets for virtual screening with machine learning. University of Hamburg. 2017.

• Daniel Zügner. Generative adversarial networks for graphs. University of Hamburg. 2017.

• Philipp Schlesinger. Learning from failure in human generated models. University of Hamburg. 2017.

• Edgar de Jesús Ek Chacón. Sistema inteligente con integración multisensorial y aprendizaje de abstracciones. Universidad Autónoma de Yucatán. 2017.

• Miguel Córdova Aké. Detección y clasificación de ojos abiertos y cerrados usando aprendizaje supervisado. Universidad Autónoma de Yucatán. 2015.

• Erberth Jesús Castillo Ceh. Algoritmo de segmentación de objetos basado en la forma con aplicación en sistemas de vigilancia. Universidad Autónoma de Yucatán. 2015.

• Roger David Soberanis Mukul. Algoritmos de segmentación de Trypanosoma cruzi en imágenes de muestras sanguíneas. Universidad Autónoma de Yucatán. 2014.

• Amir Gilberto Santos Kú. Arquitectura actor-actor-critico en brazos robot. Universidad Autónoma de Yucatán. 2014.

• César Iván Cobos May. Algoritmos para diseñar plantillas de convolución que determinan características tipo Haar. Universidad Autónoma de Yucatán. 2014.

• Daniel G Cantón Puerto. Selección de servicios Web utilizando modelos ocultos de Markov. Universidad Autónoma de Yucatán. 2014.

• Francisco de Jesús Coral Sabido. Método actor-crítico para tareas de aprendizaje por refuerzo en espacios de estados y una acción con parámetros continuos. Universidad Autónoma de Yucatán. 2013.

• Martha Varguez Moo. Diseño e implementación de técnicas de aprendizaje automático para el descubrimiento de servicios web semánticos. Universidad Autónoma de Yucatán. 2012.

• Luis Blanco Cocom. Modelación del desempeño de catalizadores basados en Pt en la reacción de oxidación de etanol a través de la teoría del funcional de la densidad y algoritmos genéticos. Universidad Autónoma de Yucatán. 2012.

• Diego Campos Sobrino. Algoritmos de aprendizaje automático para la auto-localización robótica con visión omnidireccional. Universidad Autónoma de Yucatán. 2012.

Bachelor's Theses

• Oscar Raul Navarrete Parra. Un enfoque de aprendizaje reforzado para optimizar la generación de diálogos en los modelos neuronales de arquitectura transformer. 2023.

• Miguel Ángel Quiñonez Ramírez. Mobile robot indoor navigation using deep reinforcement learning. Universidad Autónoma de Yucatán. 2022.

• Ismael de Jesús Ávila Uc. Red neuronal convolucional para detección y clasificación de señales de tránsito. 2020.

• Roger Soberanis Mukul. Detección de Trypanosoma cruzi en imágenes obtenidas a partir de muestras sanguíneas. 2012.

• Amir Gilberto Santos Kú. Redes Bayesianas para la implementación de protocolos de evaluación psicológica, forense y medida de intervención biopsicosocial para menores y adolescentes en procesos jurídicos. Universidad Autónoma de Yucatán. 2011.

• Martha Varguez Moo. Árboles de decisión para la implementación de protocolos de evaluación psicológica, forense y medida de intervención biopsicosocial para menores y adolescentes en procesos jurídicos. Universidad Autónoma de Yucatán. 2010.

• Miguel Ángel Arce Cambranis. Un algoritmo de enrutamiento para redes inalámbricas ad-hoc basado en agentes con la arquitectura sub-sumption. Universidad Autónoma de Yucatán. 2005.

Research interests

Deep neural networks, reinforcement learning and natural language processing:

1. Deep neural networks. I am mainly interested in object recognition, frame prediction and image synthesis using convolutional networks, generative adversarial networks, and transformer networks. State-of-the-art deep neural networks are doing things that were simply impossible to accomplish just a couple of years ago. I am investigating the properties of the most recently proposed deep models and how they can be used to develop novel applications.

2. Reinforcement learning and natural language processing. I am interested on natural language learning through  reinforcement learning and transformer neural networks. Computer programs capable to read text and answer relevant specific questions have never been more interesting than today. I am investigating robust algorithms for programming such systems to be used in self-training conversational bots.

Peer reviewer of journals

1. ACM Computing Surveys.

2. Journal of Ambient Intelligence and Humanized Computing.

3. Artificial Intelligence Review.

4. Expert Systems with Applications.

5. IEEE Transactions on Cognitive and Developmental Systems.

6. Scientific Reports.

Research projects

Project 1. Image-based reinforcement learning using vision transformers

We investigate the compact state representation problem in image-based reinforcement learning using vision transformers. Vision transformers are thought to surpass the performance of convolutional neural networks in the computer vision domain, in the same way that standard transformers have surpassed the performance of recurrent neural networks in the natural language processing domain. The use of vision transformers in reinforcement learning requires the pre-training of the transformer in order to accelerate the control learning process. With this project we aim to find an optimal way to pre-train or accelerate the learning process of a visual transformer that will serve as the state representation module of an image-based reinforcement learning agent.

Project 2. Next generation self-training conversational systems: leveraging neural language models, knowledge bases and reinforcement learning

Conversational systems are often heuristically-driven and thus the flow of conversation as well as the capabilities are specifically tailored to a single application. Application-specific rule-based systems can achieve reasonably good performance due to the incorporation of expert domain knowledge. However, due to their limitations when they need to be updated with new knowledge and rules, there are ongoing efforts to use data-driven or statistical conversational systems based on reinforcement learning. In theory, these data-driven conversational systems are capable of self-adapting based on interactions with real users. Additionally, they require less development effort but at a cost of significant learning time. Although very promising they still need to overcome several limitations before they are adopted for real-world applications.

In recent years, state of the art neural language models such as BERT and GPT have remarkably improved the performance of language processing systems. These transformer neural architectures trained with large portions of the information available in the world wide web can handle several languages and are being used in applications such as language translation, chatbots and text summarization, to mention a few. At the same time, an increasing number of researchers have explored the use of reinforcement learning algorithms as key components in the solution of various natural language processing tasks. For instance, conversational systems capable to read text and answer relevant specific questions as part of a dialogue are of increasing interest nowadays, both in industry and academia.

This project aims to investigate robust algorithms to be used in the next generation of self-training conversational systems, leveraging neural language models, expert knowledge bases and reinforcement learning.

Publications

Web of Science Google Scholar

Books

1. Anabel Martin-Gonzalez, Victor Uc-Cetina (Eds). Intelligent Computing Systems: First International Symposium ISICS 2016. Communications in Computer and Information Science. Springer. ISSN: 1865-0929. 2016.

Journals

1. Laura Álvarez-González, Víctor Uc-Cetina, Anabel Martin-González. Generative Adversarial Networks for Data Augmentation in Person Re-Identification. Submitted toThe Visual Computer. 2023.

2. Víctor Uc-Cetina. Recent Advances in Software Effort Estimation using Machine Learning. 2023. [arXiv version]

3. Miguel Quiñones-Ramírez, Jorge Ríos-Martínez, Víctor Uc-Cetina. Robot path planning using deep reinforcement learning. Submitted to IJPRAI. 2023. [arXiv version]

4. Oscar R. Navarrete-Parra, Víctor Uc-Cetina, Jorge Reyes-Magaña. Aligning a medium-size GPT model in English to a small closed domain in Spanish using reinforcement learning. PLN. 2023. [Journal version]. 2023.

5. Víctor Uc-Cetina, Laura Álvarez-González, Anabel Martin-González. A review on generative adversarial networks for data augmentation in person re-identification systems. Abstraction and Application, Vol. 39, pp. 101-111. ISSN: 2007-2635. 2023. [arXiv version]

6. D. J. Ruz-Suarez, A. Martin-Gonzalez, C. Brito-Loeza, V. Uc-Cetina. Fully-convolutional neural networks ensemble for comet segmentation in single cell gel electrophoresis assay images. Computer Methods in Biomechanics and Biomedical Engineering: Imaging & Visualization [Journal version]. 2023.

7. Víctor Uc-Cetina, Nicolás Navarro-Guerrero, Anabel Martin-Gonzalez, Cornelius Weber, Stefan Wermter. Survey on reinforcement learning for language processing. Artificial Intelligence Review. Electronic ISSN: 1573-7462,  print ISSN: 0269-2821. 2022. [pdf]

8. Allan Ojeda-Pat, Anabel Martín-González, Víctor Uc-Cetina. Revisión de métodos de aprendizaje automático para detectar al parásito de la enfermedad de Chagas (Machine learning methods review to detect Chagas disease parasite). Revista Investigación y Ciencia de la Universidad Autónoma de Aguascalientes. ISSN  2521-9758. 2020.

9. Saúl H. Martínez-Treviño, Víctor Uc-Cetina, María A. Fernández-Herrera, and Gabriel Merino. Prediction of Natural Products Classes Using Machine Learning and 13C NMR Spectroscopic Data. Journal of Chemical Information and Modeling. ISSN: 1549-960X.  2020.

10. Jared D. T. Guerrero-Sosa, Víctor Hugo Menéndez-Domínguez, Víctor Uc-Cetina. Redes neuronales recurrentes y su uso para la conversión de cadenas. Komputer Sapiens, Año XI, Vol. II, pp. 51-55. ISSN: 2007-0691. 2019.

11. Diego Campos Sobrino, Mario Campos Soberanis, Iván Martínez Chin, Víctor Uc Cetina. Corrección de errores del reconocedor de voz de Google usando métricas de distancia fonética. Research in Computing Science, Vol. 148(1), pp. 57-70. ISSN: 1870-4069. 2019.

12. Juan José Negrón-Granados, Ricardo Legarda Sáenz, Víctor Uc Cetina. Enfoque de la clasificación de vegetación polinífera usando imágenes multiespectrales y redes neuronales. Research in Computing Science, Vol. 147(4), pp. 79-91. ISSN: 1870-4069. 2018.

13. Víctor Uc Cetina. La singularidad tecnológica: cuando las máquinas sean más inteligentes que los humanos. Abstraction and Application, Vol. 19, pp. 1-5. ISSN: 2007-2635. 2018.

14. J. Jimenez, A. Martin, V. Uc, A. Espinosa. Mexican sign language alphanumerical gestures recognition using 3d Haar-like features. IEEE Latin America Transactions, Vol. 15, No. 10, pp. 2000-2005. ISSN: 1548-0992. 2017.

15. José L. Medina-Catzin, Anabel Martin-Gonzalez, Carlos Brito-Loeza, Victor Uc-Cetina. Body gestures recognition system to control a service robot. International Journal of Information Technology and Computer Science, Vol. 9, No. 9, pp. 69-76. ISSN: 2074-9007. 2017.

16. Daniel G Canton-Puerto, Francisco Moo-Mena, Víctor Uc-Cetina. QoS-Based Web Services Selection Using a Hidden Markov Model. Journal of Computers, Vol. 12, No. 1, pp. 48-56. ISSN: 1796-203X. 2017.

17. Francisco Moo-Mena, Rafael Hernández Ucán, Víctor Uc-Cetina, Francisco Madera Ramírez. Web service composition using the bidirectional Dijkstra algorithm. IEEE Latin America Transactions, Vol 14, No. 5, pp. 2522-2528. ISSN: 1548-0992. 2016.

18. Carlos Brito-Loeza, Ke Chen, Víctor Uc-Cetina. Image denoising using the Gaussian curvature of the image surface. Numerical Methods for Partial Differential Equations, Vol. 32, No. 3, pp. 1066-1089. ISSN: 1098-2426. 2016.

19. Anabel Martin-Gonzalez, A. Chi-Poot, Víctor Uc-Cetina. Usability evaluation of an augmented reality system for teaching Euclidean vectors. Innovations in Education and Teaching International, Vol. 53, No. 6, pp. 627-636. ISSN: 1470-3300. 2015.

20. Victor Uc-Cetina, Carlos Brito-Loeza, y Hugo Ruiz-Piña. Chagas Parasite Detection in Blood Images using AdaBoost. Computational and Mathematical Methods in Medicine (2015). Vol. 2015, No. 1, pp. 1-13, Article ID 139681. ISSN: 1748-670X. 2015.

21. Cesar Cobos-May, Victor Uc-Cetina, Carlos Brito-Loeza, y Anabel Martin-Gonzalez. A Convex Set Based Algorithm to Automatically Generate Haar-Like Features. Computer Science and Applications (2015). Vol. 2, No. 2, pp. 64-70. ISSN: 2333-908X. 2015.

22. Victor Uc-Cetina, Francisco Moo-Mena, y Rafael Hernandez-Ucan. Composition of Web Services Using Markov Decision Processes and Dynamic Programming. The Scientific World Journal (2015). Vol. 2015, No. 1, pp. 1-9, Article ID 545308. ISSN: 1537-744X. 2015.

23. Carlos Brito-Loeza, Víctor Uc-Cetina, Anabel Martin-Gonzalez. Introducción a los Métodos Variacionales en Procesamiento de Imágenes: Filtros de Ruido. Abstraction and Application, Vol. 10, pp. 19-34. ISSN: 2007-2635. 2014.

24. Jorge Lugo-Jiménez, Víctor Uc-Cetina. Geometría Fractal en la Superficie Lunar. Abstraction and Application, Vol. 9, pp. 19-26. ISSN: 2007-2635. 2013.

25. Roger Soberanis-Mukul, Víctor Uc-Cetina, Carlos Brito-Loeza, Hugo Ruiz-Piña. An Automatic Algorithm for the Detection of Trypanosoma cruzi Parasites in Blood Sample Images. Computer Methods and Programs in Biomedicine, Vol. 9, pp. 19-26. ISSN: 0169-2607. 2013.

26. Víctor Uc-Cetina, Carlos Brito-Loeza, Hugo Ruiz-Piña. Chagas Parasites Detection through Gaussian Discriminant Analysis. Abstraction and Application, Vol. 8, pp. 6-17. ISSN: 2007-2635. 2013.

27. Víctor Uc-Cetina, A Novel Reinforcement Learning Architecture for Continuous State and Action Spaces. Advances in Artificial Intelligence, Vol. 2013, Article ID 492852, 10 pages. doi:10.1155/2013/492852, ISSN: 1687-7489. 2013.

28. Martha Varguez-Moo, Francisco Moo-Mena, Víctor Uc-Cetina. Use of Classification Algorithms for Semantic Web Services Discovery. Journal of Computers, Vol. 8, No. 7, pp. 1810-1814. ISSN: 1796-203X. 2013.

29. Martha Varguez-Moo, Victor Uc-Cetina, Carlos Brito-Loeza. Clasificación de Documentos usando Máquinas de Vectores de Apoyo. Abstraction and Application, Vol. 6, pp. 40-51. ISSN: 2007-2635. 2012.

Conferences

1. Jesús Cabrera González, Anabel Martin-gonzalez, Jorge Lugo-Jiménez, Víctor Uc-Cetina. Towards and Automatic Counter of Lunar Craters. Proceedings of the CCE. 2014.

2. José Emiliano López-Noriega, Miguel Iván Fernández-Valladares, Víctor Uc-Cetina. Glove-Based Sign Language Recognition Solution to Assist Communication for Deaf Users. Proceedings of the CCE. 2014.

3. César Iván Cobos-May, Víctor Uc-Cetina, Carlos Brito-Loeza. Algoritmos para Determinar Características de Tipo Haar. Proceedings of the CONIEEM. ISSN 1665-0271. 2013.

4. Roger Soberanis-Mukul, Víctor Uc-Cetina, Carlos Brito-Loeza, Hugo Ruiz-Piña. Detección de Trypanosoma cruzi en Imágenes Obtenidas a partir de Muestras Sanguíneas. Proceedings of the CONIEEM. ISSN 1665-0271. 2013.

5. Francisco de Jesús Coral-Sabido, Carlos Brito-Loeza, Víctor Uc-Cetina. Seguimiento de rutas utilizando aprendizaje por refuerzo en espacios de estado y acción continuos. Proceedings of the CONIEEM. ISSN 1665-0271. 2012.

6. Daniel G. Cantón-Puerto, Víctor Uc-Cetina, Francisco Moo-Mena. Dynamic Web Services Selection using a Hidden Markov Model. Proceedings of the CCE. 2012.

7. Diego Campos-Sobrino, Francisco Coral-Sabido, Martha Varguez-Moo, Victor Uc-Cetina, Arturo Espinosa-Romero. Mobile Robot Self-Localization based on Omnidirectional Vision and Gaussian Models. Proceedings of the Tenth International Conference on Machine Learning and Applications (ICMLA'11). Honolulu, Hawaii, United States. 2011.

8. Francisco de Jesús Coral Sabido, Carlos Brito Loeza, Víctor Uc Cetina. Seguimiento de rutas utilizando aprendizaje por refuerzo en espacios de estado y acción continuos. CONIEEM. ISSN 1665-0271. 2012.

9. Daniel G. Cantón Puerto, Victor Uc Cetina, Francisco Moo Mena. Dynamic Web Services Selection using a Hidden Markov Model. Proceedings of the CCE. 2012.

10. Diego Campos-Sobrino, Francisco Coral-Sabido, Martha Varguez-Moo, Victor Uc-Cetina y Arturo Espinosa-Romero. Mobile Robot Self-Localization based on Omnidirectional Vision and Gaussian Models. Proceedings of the Tenth International Conference on Machine Learning and Applications (ICMLA'11). Honolulu, Hawaii, United States. 2011.

11. Martha Varguez-Moo, Francisco Moo-Mena, Victor Uc-Cetina. Toward Semantic Web Services Discovery: a Machine Learning Approach. Book chapter: Avances en Informática y Sistemas Computacionales. ISBN 978-607-606-039-1. 2011.

12. Luis Daniel Blanco Cocom, Luis Carlos Ordónez López, Eric José Ávila Vales, Victor Emanuel Uc Cetina. Modelación del desempeño de catalizadores basados en PT en la reacción de oxidación de etanol a través de la teoría del funcional de la densidad. Proceedings of the XLIV Congreso Nacional de la Sociedad Matematica Mexicana. San Luis Potosi, Mexico. 2011.

13. X. Sierra-Canto, F. Madera-Ramírez, V. Uc-Cetina. Parallel Training of a Back-Propagation Neural Network using CUDA. In Proceedings of the 9th International Conference on Machine Learning and Applications (ICMLA'10). Washington DC, USA. 2010.

14. V. Uc Cetina, H. D. Burkhard. Robot Control Learning in Continuous State and Action Spaces using 2 Actors and 1 Critic. In Proceedings of the First Workshop on Learning and Data Mining for Robotics (LEMIR'09) of the European Conference on Machine Learning (ECML'09). Bled, Slovenia. 2009.

15. V. Uc Cetina. Learning like Humans do: from Examples and Self-Generated Experiences. In Proceedings of the 6th International Conference on Human System Learning (ICHSL'08). Toulouse, France. 2008.

16. V. Uc Cetina. Multilayer Perceptrons with Radial Basis Functions as Value Functions in Reinforcement Learning. In Proceedings of the 16th European Symposium on Artificial Neural Networks (ESANN'08). Bruges, Belgium. 2008.

17. V. Uc Cetina. Autonomous Agent Learning using an Actor-Critic Algorithm and Behavior Models (short paper). In Proceedings of the 7th International Joint Conference on Autonomous Agents and Multi-Agent Systems (AAMAS'08). Estoril, Portugal. 2008.

18. V. Uc Cetina. Supervised Reinforcement Learning Using Behavior Models. In Proceedings of the 6th International Conference on Machine Learning and Applications (ICMLA'07). Cincinnati, Ohio, USA. 2007.

19. V. Uc Cetina. A Multiagent Architecture for Concurrent Reinforcement Learning. In Proceedings of the 14th European Symposium on Artificial Neural Networks (ESANN'06). Bruges, Belgium. 2006.

20. V. Uc Cetina, H. Terashima Marín. Container Loading Optimization Using a Genetic Algorithm. In Proceedings of the Workshop on Soft Computing of the 2nd Mexican International Conference on Artificial Intelligence (MICAI'02). Merida, Mexico. 2002.