Estudio de la reducción del sobreajuste en arquitecturas de redes neuronales residuales ResNet en un escenario de clasificación de patrones

dc.contributor.advisorRiaño Rojas, Juan Carlos
dc.contributor.advisorGallego Restrepo, Fernando Andrés
dc.contributor.authorChacón Chamorro, Manuela Viviana
dc.contributor.cvlacChacón Chamorro, Manuela [0000166834]spa
dc.contributor.researchgroupPcm Computational Applicationsspa
dc.date.accessioned2023-07-18T19:29:38Z
dc.date.available2023-07-18T19:29:38Z
dc.date.issued2023
dc.descriptiongraficas, tablasspa
dc.description.abstractLas redes neuronales artificiales son una técnica de aprendizaje automático inspirada en el funcionamiento biológico de las neuronas, actualmente soportan gran parte de la denominada Inteligencia Artificial. Pese a su notable evolución estos algoritmos presentan el problema de sobreajuste, "memorización de los datos de entrenamiento", lo cual disminuye la capacidad de generalización. En este trabajo se estudió el sobreajuste en un escenario de clasificación de patrones y se determinó un método para resolver el problema. Este estudio se realizó para la arquitectura de neuronal residual (ResNet) y se sustentó en el análisis de las propiedades matemáticas de la función que representa esta estructura, en particular, la continuidad de Lipschitz. La validación del método se realizó comparando su desempeño con las técnicas convencionales de reducción de sobreajuste: la regularización L1, L2 y Dropout. Variando la profundidad de la red se realizaron dos experimentos de clasificación con los conjuntos de datos Digits y Fashion de MNIST. También se efectuaron pruebas en arquitecturas definidas para 3 conjuntos de datos convencionales y 3 de datos sintéticos. Adicionalmente, se realizaron dos experimentos que incluyeron imágenes adversarias. El método desarrollado presenta un desempeño destacable logrando: comportamiento similar en las curvas de aprendizaje para entrenamiento y prueba, menor variabilidad del modelo al cambiar el conjunto de entrenamiento, reducción de la cota de Lipschitz, tolerancia a las pruebas adversarias. En síntesis, el método propuesto resultó idóneo en la reducción del sobreajuste en las arquitecturas residuales de los experimentos y tolera de manera sobresaliente ataques adversarios. (Texto tomado de la fuente)spa
dc.description.abstractArtificial neural networks are a technique of machine learning inspired by the biological functioning of neurons, currently supporting a significant portion of the so-called Artificial Intelligence. Despite their notable evolution, these algorithms present the problem of overfitting, "training data memorization", which reduces the capacity of generalization. In this work, overfitting in a pattern classification scenario was studied and a method to solve the problem was determined. This study was carried out for the Residual Neural Network architecture (ResNet) and was based on the analysis of the mathematical properties of the function that represents this structure, in particular, the Lipschitz continuity. The method was validated by comparing its performance with conventional overfitting reduction techniques: L1, L2 and Dropout regularization. Varying the depth of the network, two classification experiments were performed with the data sets Digits and Fashion MNIST. Tests were also performed on architectures defined for 3 conventional data sets and 3 synthetic data sets. Additionally, two experiments were conducted that included adversarial images. The developed method posed remarkable performance achieving: similar behavior in the learning curves for train and test set, less variability of the model when changing the train set, reduction of the Lipschitz bound and adversarial test tolerance. In summary, the method is suitable to reduce overfitting in residual architectures of the experiments and it tolerates adversary attacks in an outstanding way.eng
dc.description.curricularareaMatemáticas Y Estadística.Sede Manizalesspa
dc.description.degreelevelMaestríaspa
dc.description.degreenameMagíster en Ciencias - Matemática Aplicadaspa
dc.format.extentxx, 150 páginasspa
dc.format.mimetypeapplication/pdfspa
dc.identifier.instnameUniversidad Nacional de Colombiaspa
dc.identifier.reponameRepositorio Institucional Universidad Nacional de Colombiaspa
dc.identifier.repourlhttps://repositorio.unal.edu.co/spa
dc.identifier.urihttps://repositorio.unal.edu.co/handle/unal/84211
dc.language.isospaspa
dc.publisherUniversidad Nacional de Colombiaspa
dc.publisher.branchUniversidad Nacional de Colombia - Sede Manizalesspa
dc.publisher.facultyFacultad de Ciencias Exactas y Naturalesspa
dc.publisher.programManizales - Ciencias Exactas y Naturales - Maestría en Ciencias - Matemática Aplicadaspa
dc.relation.referencesC. C. Aggarwal, Neural Networks and Deep Learning. Springer, 2018.spa
dc.relation.referencesZ.-Q. Zhao, P. Zheng, S.-t. Xu, and X. Wu, “Object detection with deep learning: A review,” 2018. [Online]. Available: https://arxiv.org/abs/1807.05511spa
dc.relation.referencesA. Kamilaris and F. X. Prenafeta-Bold ́u, “A review of the use of convolutional neural networks in agriculture,” The Journal of Agricultural Science, vol. 156, no. 3, p.312–322, 2018. [Online]. Available: https://doi.org/10.1017/S0021859618000436spa
dc.relation.referencesA. Creswell, T. White, V. Dumoulin, K. Arulkumaran, B. Sengupta, and A. A. Bharath, “Generative adversarial networks: An overview,” IEEE Signal Processing Magazine, vol. 35, no. 1, pp. 53–65, jan 2018. [Online]. Available: https://doi.org/10.1109%2Fmsp.2017.2765202spa
dc.relation.referencesA. Fadaeddini, M. Eshghi, and B. Majidi, “A deep residual neural network for low altitude remote sensing image classification,” in 2018 6th Iranian Joint Congress on Fuzzy and Intelligent Systems (CFIS), 2018, pp. 43–46. [Online]. Available: https://ieeexplore.ieee.org/document/8336623spa
dc.relation.referencesL. Massidda, M. Marrocu, and S. Manca, “Non-intrusive load disaggregation by convolutional neural network and multilabel classification,” Applied Sciences, vol. 10, no. 4, 2020. [Online]. Available: https://www.mdpi.com/2076-3417/10/4/1454spa
dc.relation.referencesK. Muralitharan, R. Sakthivel, and R. Vishnuvarthan, “Neural network based optimization approach for energy demand prediction in smart grid,” Neurocomputing, vol. 273, pp. 199–208, 2018. [Online]. Available: https://www.sciencedirect.com/science/article/pii/S0925231217313681spa
dc.relation.referencesO. Al-Salman, J. Mustafina, and G. Shahoodh, “A systematic review of artificial neural networks in medical science and applications,” in 2020 13th International Conference on Developments in eSystems Engineering (DeSE), 2020, pp. 279–282. [Online]. Available: https://ieeexplore.ieee.org/document/9450245spa
dc.relation.referencesM. M. Bejani and M. Ghatee, “A systematic review on overfitting control in shallow and deep neural networks,” Artificial Intelligence Review, vol. 54, no. 8, pp. 6391–6438, 2021. [Online]. Available: https://doi.org/10.1007/s10462-021-09975-1spa
dc.relation.referencesS. Salman and X. Liu, “Overfitting mechanism and avoidance in deep neural networks,” arXiv preprint arXiv:1901.06566, 2019. [Online]. Available: https://arxiv.org/abs/1901.06566spa
dc.relation.referencesX. Ying, “An overview of overfitting and its solutions,” in Journal of physics: Conference series, vol. 1168, no. 2. IOP Publishing, 2019, p. 022022. [Online]. Available: https://iopscience.iop.org/article/10.1088/1742-6596/1168/2/022022spa
dc.relation.referencesI. Bilbao and J. Bilbao, “Overfitting problem and the over-training in the era of data: Particularly for artificial neural networks,” in 2017 eighth international conference on intelligent computing and information systems (ICICIS). IEEE, 2017, pp. 173–177. [Online]. Available: https://ieeexplore.ieee.org/abstract/document/8260032spa
dc.relation.referencesR. T. Chen, Y. Rubanova, J. Bettencourt, and D. K. Duvenaud, “Neural ordinary differential equations,” Advances in neural information processing systems, vol. 31, 2018. [Online]. Available: https://proceedings.neurips.cc/paper/2018/file/69386f6bb1dfed68692a24c8686939b9-Paper.pdfspa
dc.relation.referencesD. Karlsson and O. Svanstr ̈om, “Modelling dynamical systems using neural ordinary differential equations,” Master’s thesis, Chalmers University of Technology, 2019. [Online]. Available: https://odr.chalmers.se/server/api/core/bitstreams/83a4c17f-35e7-43ce-ac60-43a0b799f82f/contentspa
dc.relation.referencesE. Weinan, “A proposal on machine learning via dynamical systems,” Communications in Mathematics and Statistics, vol. 1, no. 5, pp. 1–11, 2017. [Online]. Available: https://link.springer.com/article/10.1007/s40304-017-0103-zspa
dc.relation.referencesM. Benning, E. Celledoni, M. J. Ehrhardt, B. Owren, and C.-B. Schonlieb, “Deep learning as optimal control problems: Models and numerical methods,” arXiv preprint arXiv:1904.05657, 2019. [Online]. Available: https://arxiv.org/abs/1904.05657spa
dc.relation.referencesQ. Li, L. Chen, C. Tai et al., “Maximum principle based algorithms for deep learning,” Journal of Machine Learning Research, pp. 1–29, 2018. [Online]. Available: https://www.jmlr.org/papers/volume18/17-653/17-653.pdfspa
dc.relation.referencesQ. Li and S. Hao, “An optimal control approach to deep learning and applications to discrete-weight neural networks,” in International Conference on Machine Learning. PMLR, 2018, pp. 2985–2994. [Online]. Available: http://proceedings.mlr.press/v80/li18b/li18b.pdfspa
dc.relation.referencesJ. Han, Q. Li et al., “A mean-field optimal control formulation of deep learning,” Research in the Mathematical Sciences, vol. 6, no. 1, pp. 1–41, 2019. [Online]. Available: https://link.springer.com/article/10.1007/s40687-018-0172-yspa
dc.relation.referencesE. Haber and L. Ruthotto, “Stable architectures for deep neural networks,” Inverse problems, vol. 34, no. 1, p. 014004, 2017. [Online]. Available: https://iopscience.iop.org/article/10.1088/1361-6420/aa9a90/metaspa
dc.relation.referencesB. Chang, L. Meng, E. Haber, F. Tung, and D. Begert, “Multi-level residual networks from dynamical systems view,” arXiv preprint arXiv:1710.10348, 2017. [Online]. Available: https://arxiv.org/abs/1710.10348spa
dc.relation.referencesM. Ciccone, M. Gallieri, J. Masci, C. Osendorfer, and F. Gomez, “Nais-net: Stable deep networks from non-autonomous differential equations,” Advances in Neural Information Processing Systems, vol. 31, 2018. [Online]. Available: https://proceedings.neurips.cc/paper/2018/file/7bd28f15a49d5e5848d6ec70e584e625-Paper.pdfspa
dc.relation.referencesC. Finlay, J. Calder, B. Abbasi, and A. Oberman, “Lipschitz regularized deep neural networks generalize and are adversarially robust,” arXiv preprint arXiv:1808.09540, 2018. [Online]. Available: https://arxiv.org/abs/1808.09540spa
dc.relation.referencesP. Pauli, A. Koch, J. Berberich, P. Kohler, and F. Allg ̈ower, “Training robust neural networks using lipschitz bounds,” IEEE Control Systems Letters, vol. 6, pp. 121–126, 2021. [Online]. Available: https://ieeexplore.ieee.org/document/9319198spa
dc.relation.referencesH. Gouk, E. Frank, B. Pfahringer, and M. J. Cree, “Regularisation of neural networks by enforcing lipschitz continuity,” Machine Learning, vol. 110, no. 2, pp. 393–416, 2021. [Online]. Available: https://link.springer.com/article/10.1007/s10994-020-05929-wspa
dc.relation.referencesB. Dherin, M. Munn, M. Rosca, and D. G. Barrett, “Why neural networks find simple solutions: the many regularizers of geometric complexity,” arXiv preprint arXiv:2209.13083, 2022. [Online]. Available: https://arxiv.org/abs/2209.13083spa
dc.relation.referencesT. Zhou, Q. Li, H. Lu, Q. Cheng, and X. Zhang, “Gan review: Models and medical image fusion applications,” Information Fusion, vol. 91, pp. 134–148, 2023. [Online]. Available: https://doi.org/10.1016/j.inffus.2022.10.017spa
dc.relation.referencesC. A. Charu, Neural networks and deep learning: a textbook. Spinger, 2018.spa
dc.relation.referencesS. Theodoridis, “Neural networks and deep learning,” Machine Learning, pp. 875–936, 2015spa
dc.relation.referencesF. Rosenblatt, “The perceptron: a probabilistic model for information storage and organization in the brain.” Psychological review, vol. 65, no. 6, p. 386, 1958. [Online]. Available: https://psycnet.apa.org/record/1959-09865-001spa
dc.relation.referencesB. Pang, E. Nijkamp, and Y. N. Wu, “Deep learning with tensorflow: A review,” Journal of Educational and Behavioral Statistics, vol. 45, no. 2, pp. 227–248, 2020. [Online]. Available: https://doi.org/10.3102/1076998619872761spa
dc.relation.referencesK. He, X. Zhang, S. Ren, and J. Sun, “Deep residual learning for image recognition,” in Proceedings of the IEEE conference on computer vision and pattern recognition, 2016, pp. 770–778. [Online]. Available: https://ieeexplore.ieee.org/document/7780459spa
dc.relation.referencesD. Hunter, H. Yu, M. S. Pukish III, J. Kolbusz, and B. M. Wilamowski, “Selection of proper neural network sizes and architectures a comparative study,” IEEE Transactions on Industrial Informatics, vol. 8, no. 2, pp. 228–240, 2012. [Online]. Available: https://ieeexplore.ieee.org/document/6152147spa
dc.relation.referencesMATLAB, “Statistics and machine learning toolbox.” [Online]. Available: https://la.mathworks.com/products/statistics.htmlspa
dc.relation.referencesR, “neuralnet: Training of neural networks.” [Online]. Available: https://www.rdocumentation.org/packages/neuralnet/versions/1.44.2/topics/neuralnetspa
dc.relation.referencesTensorFlow, “Tensorflow 2.10.0.” [Online]. Available: https://www.tensorflow.org/spa
dc.relation.referencesJ. Reunanen, “Overfitting in making comparisons between variable selection methods,” Journal of Machine Learning Research, vol. 3, pp. 1371–1382, 2003. [Online]. Available: https://www.jmlr.org/papers/volume3/reunanen03a/reunanen03a.pdfspa
dc.relation.referencesI. Guyon and A. Elisseeff, “An introduction to variable and feature selection,” Journal of machine learning research, vol. 3, pp. 1157–1182, 2003. [Online]. Available: https://www.jmlr.org/papers/volume3/guyon03a/guyon03a.pdfspa
dc.relation.referencesF. Johnson, A. Valderrama, C. Valle, B. Crawford, R. Soto, and R. Nanculef, “Automating configuration of convolutional neural network hyperparameters using genetic algorithm,” IEEE Access, vol. 8, pp. 156 139–156 152, 2020. [Online]. Available: https://ieeexplore.ieee.org/document/9177040spa
dc.relation.referencesY. Zhu, G. Li, R. Wang, S. Tang, H. Su, and K. Cao, “Intelligent fault diagnosis of hydraulic piston pump combining improved lenet-5 and pso hyperparameter optimization,” Applied Acoustics, vol. 183, p. 108336, 2021. [Online]. Available: https://www.sciencedirect.com/science/article/abs/pii/S0003682X21004308spa
dc.relation.referencesA. Gaspar, D. Oliva, E. Cuevas, D. Zaldıvar, M. Pérez, and G. Pajares, “Hyperparameter optimization in a convolutional neural network using metaheuristic algorithms,” in Metaheuristics in Machine Learning: Theory and Applications. Springer, 2021, pp. 37–59. [Online]. Available: https://link.springer.com/chapter/10.1007/978-3-030-70542-8 2spa
dc.relation.referencesO. S. Steinholtz, “A comparative study of black-box optimization algorithms for tuning of hyper-parameters in deep neural networks,” Lulea University of Technology, 2018. [Online]. Available: https://ltu.divaportal.org/smash/get/diva2:1223709/FULLTEXT01.pdfspa
dc.relation.referencesL. Lugo, “A recurrent neural network approach for whole genome bacteria classification,” Master’s thesis, Universidad Nacional de Colombia, Bogota, Colombia, 2018.spa
dc.relation.referencesA. T. Sarmiento and O. Soto, “New product forecasting demand by using neural networks and similar product analysis.” Master’s thesis, Universidad Nacional de Colombia, Medellin, Colombia, 2014.spa
dc.relation.referencesA. E. Casas Fajardo, “Propuesta metodológica para calcular el avalúo de un predio empleando redes neuronales artificiales,” Master’s thesis, Universidad Nacional de Colombia, Bogotá, Colombia, 2014.spa
dc.relation.referencesS. Ortega Alzate, “Exploración de las redes neuronales para la proyección de la máxima pérdida esperada de una póliza de seguros: aplicación para un seguro previsionales,” Master’s thesis, Universidad Nacional de Colombia, Medellin, Colombia, 2021.spa
dc.relation.referencesD. Collazos, “Kernel-based enhancement of general stochastic network for supervised learning,” Master’s thesis, Universidad Nacional de Colombia, Manizales, Colombia, 2016.spa
dc.relation.referencesY. Lu, A. Zhong, Q. Li, and B. Dong, “Beyond finite layer neural networks: Bridging deep architectures and numerical differential equations,” in International Conference on Machine Learning. PMLR, 2018, pp. 3276–3285. [Online]. Available: http://proceedings.mlr.press/v80/lu18d/lu18d.pdfspa
dc.relation.referencesB. Geshkovski and E. Zuazua, “Turnpike in optimal control of pdes, resnets, and beyond,” Acta Numerica, vol. 31, pp. 135–263, 2022. [Online]. Available: https://doi.org/10.1017/S0962492922000046spa
dc.relation.referencesD. Ruiz-Balet and E. Zuazua, “Neural ode control for classification, approximation and transport,” arXiv preprint arXiv:2104.05278, 2021. [Online]. Available: https://arxiv.org/abs/2104.05278spa
dc.relation.referencesD. Ruiz-Balet, E. Affili, and E. Zuazua, “Interpolation and approximation via momentum resnets and neural odes,” Systems & Control Letters, vol. 162, p. 105182, 2022. [Online]. Available: https://doi.org/10.1016/j.sysconle.2022.105182spa
dc.relation.referencesM. Fazlyab, A. Robey, H. Hassani, M. Morari, and G. Pappas, “Efficient and accurate estimation of lipschitz constants for deep neural networks,” Advances in Neural Information Processing Systems, vol. 32, 2019. [Online]. Available: https://proceedings.neurips.cc/paper/2019/file/95e1533eb1b20a97777749fb94fdb944-Paper.pdfspa
dc.relation.referencesA. Xue, L. Lindemann, A. Robey, H. Hassani, G. J. Pappas, and R. Alur, “Chordal sparsity for lipschitz constant estimation of deep neural networks,” arXiv preprint arXiv:2204.00846, 2022. [Online]. Available: https://arxiv.org/abs/2204.00846spa
dc.relation.referencesL. Massidda, M. Marrocu, and S. Manca, “Non-intrusive load disaggregation by convolutional neural network and multilabel classification,” Applied Sciences, vol. 10, no. 4, p. 1454, 2020. [Online]. Available: https://www.mdpi.com/2076-3417/10/4/1454spa
dc.relation.referencesK. Muralitharan, R. Sakthivel, and R. Vishnuvarthan, “Neural network based optimization approach for energy demand prediction in smart grid,” Neurocomputing, vol. 273, pp. 199–208, 2018. [Online]. Available: https://www.sciencedirect.com/science/article/abs/pii/S0925231217313681spa
dc.relation.referencesR. Lu and S. H. Hong, “Incentive-based demand response for smart grid with reinforcement learning and deep neural network,” Applied energy, vol. 236, pp. 937–949, 2019. [Online]. Available: https://www.sciencedirect.com/science/article/abs/pii/S0306261918318798spa
dc.relation.referencesA. P. Marug ́an, F. P. G. M ́arquez, J. M. P. Perez, and D. Ruiz-Hern ́andez, “A survey of artificial neural network in wind energy systems,” Applied energy, vol. 228, pp. 1822–1836, 2018. [Online]. Available: https://doi.org/10.1016/j.apenergy.2018.07.084spa
dc.relation.referencesF. Saeed, M. A. Khan, M. Sharif, M. Mittal, L. M. Goyal, and S. Roy, “Deep neural network features fusion and selection based on pls regression with an application for crops diseases classification,” Applied Soft Computing, vol. 103, p. 107164, 2021. [Online]. Available: https://doi.org/10.1016/j.asoc.2021.107164spa
dc.relation.referencesM. Loey, A. ElSawy, and M. Afify, “Deep learning in plant diseases detection for agricultural crops: A survey,” International Journal of Service Science, Management, Engineering, and Technology (IJSSMET), vol. 11, no. 2, pp. 41–58, 2020. [Online]. Available: https://www.igi-global.com/article/deep-learning-in-plant-diseases-detection-for-agricultural-crops/248499spa
dc.relation.referencesB. Pandey, D. K. Pandey, B. P. Mishra, and W. Rhmann, “A comprehensive survey of deep learning in the field of medical imaging and medical natural language processing: Challenges and research directions,” Journal of King Saud University-Computer and Information Sciences, 2021.spa
dc.relation.referencesA. Nogales, A. J. Garcia-Tejedor, D. Monge, J. S. Vara, and C. Antón,“A survey of deep learning models in medical therapeutic areas,” Artificial Intelligence in Medicine, vol. 112, p. 102020, 2021. [Online]. Available: https://doi.org/10.1016/j.artmed.2021.102020spa
dc.relation.referencesColciencias, “Plan Nacional de CTeI para el desarrollo del sector Tecnologías de la Información TIC 2017 - 2022,,” Bogotá, Colombia, 2017spa
dc.relation.referencesRepública de Colombia, “Plan de Desarrollo Nacional 2018-2022 “Pacto por Colombia, pacto por la equidad”,” Bogotá, Colombia, 2018spa
dc.relation.referencesColciencias, “Política Nacional de Ciencia e Innovación para el Desarrollo Sostenible Libro Verde 2030,” Bogotá, Colombia, 2018.spa
dc.relation.referencesJ. C. Riaño Rojas, “Desarrollo de una metodología como soporte para la detección de enfermedades vasculares del tejido conectivo a través de imágenes capilaroscópicas,” Ph.D. dissertation, Universidad Nacional de Colombia, Bogotá, Colombia, 2010spa
dc.relation.referencesT. T. Tang, J. A. Zawaski, K. N. Francis, A. A. Qutub, and M. W. Gaber, “Image-based classification of tumor type and growth rate using machine learning: a preclinical study,” Scientific reports, vol. 9, no. 1, pp. 1–10, 2019. [Online]. Available: https://www.nature.com/articles/s41598-019-48738-5spa
dc.relation.referencesC. A. Pedraza Bonilla and L. Rodríguez Mújica, “Método para la estimación de maleza en cultivos de lechuga utilizando aprendizaje profundo e imágenes multiespectrales,” Master’s thesis, Universidad Nacional de Colombia., Bogotá, Colombia, 2016.spa
dc.relation.referencesC. Barrios Pérez, “Zonificación agroecológica para el cultivo de arroz de riego (Oryza Sativa L.) en Colombia,” Master’s thesis, Universidad Católica de Colombia., Palmira, Colombia, 2016spa
dc.relation.referencesA. F. Montenegro and C. D. Parada, “Diseño e implementación de un sistema de detección de malezas en cultivos cundiboyacenses,” Master’s thesis, Universidad Católica de Colombia., Bogotá, Colombia, 2015spa
dc.relation.referencesM. Minsky and S. Papert, “An introduction to computational geometry,” Cambridge tiass., HIT, vol. 479, p. 480, 1969.spa
dc.relation.referencesG. Cybenko, “Approximation by superpositions of a sigmoidal function,” Mathematics of control, signals and systems, vol. 2, no. 4, pp. 303–314, 1989. [Online]. Available: https://link.springer.com/article/10.1007/BF02551274spa
dc.relation.referencesD. E. Rumelhart, G. E. Hinton, and R. J. Williams, “Learning representations by back-propagating errors,” nature, vol. 323, no. 6088, pp. 533–536, 1986. [Online]. Available: https://www.nature.com/articles/323533a0spa
dc.relation.referencesY. Yu, X. Si, C. Hu, and J. Zhang, “A review of recurrent neural networks: LSTM cells and network architectures,” Neural computation, vol. 31, no. 7, pp. 1235–1270, 2019. [Online]. Available: https://ieeexplore.ieee.org/document/8737887spa
dc.relation.referencesN. Li, S. Liu, Y. Liu, S. Zhao, and M. Liu, “Neural speech synthesis with transformer network,” in Proceedings of the AAAI Conference on Artificial Intelligence, vol. 33, no. 01, 2019, pp. 6706–6713spa
dc.relation.referencesG. Parascandolo, H. Huttunen, and T. Virtanen, “Taming the waves: sine as activation function in deep neural networks,” 2017. [Online]. Available: https://openreview.net/forum?id=Sks3zF9egspa
dc.relation.referencesJ. Heredia-Juesas and J. A. Martínez-Lorenzo, “Consensus function from an Lp− norm regularization term for its use as adaptive activation functions in neural networks,” arXiv e-prints, pp. arXiv–2206, 2022. [Online]. Available: https://arxiv.org/abs/2206.15017spa
dc.relation.referencesA. D. Jagtap, Y. Shin, K. Kawaguchi, and G. E. Karniadakis, “Deep kronecker neural networks: A general framework for neural networks with adaptive activation functions,” Neurocomputing, vol. 468, pp. 165–180, 2022. [Online]. Available: https://doi.org/10.1016/j.neucom.2021.10.036spa
dc.relation.referencesD. Devikanniga, K. Vetrivel, and N. Badrinath, “Review of meta-heuristic optimization based artificial neural networks and its applications,” in Journal of Physics: Conference Series, vol. 1362, no. 1. IOP Publishing, 2019, p. 012074. [Online]. Available: https://iopscience.iop.org/article/10.1088/1742-6596/1362/1/012074/metaspa
dc.relation.referencesN. Gupta, M. Khosravy, N. Patel, S. Gupta, and G. Varshney, “Evolutionary artificial neural networks: comparative study on state-of-the-art optimizers,” in Frontier applications of nature inspired computation. Springer, 2020, pp. 302–318. [Online]. Available: https://link.springer.com/chapter/10.1007/978-981-15-2133-1 14spa
dc.relation.referencesJ. Duchi, E. Hazan, and Y. Singer, “Adaptive subgradient methods for online learning and stochastic optimization.” Journal of machine learning research, vol. 12, no. 7, 2011. [Online]. Available: https://jmlr.org/papers/volume12/duchi11a/duchi11a.pdfspa
dc.relation.referencesD. P. Kingma and J. Ba, “Adam: A method for stochastic optimization,” arXiv preprint arXiv:1412.6980, 2014. [Online]. Available: https://arxiv.org/abs/1412.6980spa
dc.relation.referencesM. D. Zeiler, “ADADELTA: an adaptive learning rate method,” arXiv preprint arXiv:1212.5701, 2012. [Online]. Available: https://arxiv.org/abs/1212.5701spa
dc.relation.referencesS. Ruder, “An overview of gradient descent optimization algorithms,” arXiv preprint arXiv:1609.04747, 2016. [Online]. Available: https://arxiv.org/abs/1609.04747spa
dc.relation.referencesP. Netrapalli, “Stochastic gradient descent and its variants in machine learning,” Journal of the Indian Institute of Science, vol. 99, no. 2, pp. 201–213, 2019. [Online]. Available: https://link.springer.com/article/10.1007/s41745-019-0098-4spa
dc.relation.referencesS. Lawrence, C. L. Giles, and A. C. Tsoi, “Lessons in neural network training: Overfitting may be harder than expected,” in Proceedings of the Fourteenth National Conference on Artificial Intelligence, 1997, pp. 540–545. [Online]. Available: https://clgiles.ist.psu.edu/papers/AAAI-97.overfitting.hard_to_do.pdfspa
dc.relation.referencesX.-x. Wu and J.-g. Liu, “A new early stopping algorithm for improving neural network generalization,” in 2009 Second International Conference on Intelligent Computation Technology and Automation, vol. 1. IEEE, 2009, pp. 15–18. [Online]. Available: https://ieeexplore.ieee.org/document/5287721spa
dc.relation.referencesH. Liang, S. Zhang, J. Sun, X. He, W. Huang, K. Zhuang, and Z. Li, “Darts+: Improved differentiable architecture search with early stopping,” arXiv preprint arXiv:1909.06035, 2019. [Online]. Available: https://arxiv.org/abs/1909.06035spa
dc.relation.referencesL. Prechelt, “Early stopping-but when?” in Neural Networks: Tricks of the trade. Springer, 1998, pp. 55–69. [Online]. Available: https://link.springer.com/chapter/10.1007/978-3-642-35289-8 5spa
dc.relation.referencesM. Mahsereci, L. Balles, C. Lassner, and P. Hennig, “Early stopping without a validation set,” arXiv preprint arXiv:1703.09580, 2017. [Online]. Available: https://arxiv.org/abs/1703.09580spa
dc.relation.referencesC. Shorten and T. M. Khoshgoftaar, “A survey on image data augmentation for deep learning,” Journal of big data, vol. 6, no. 1, pp. 1–48, 2019. [Online]. Available: https://journalofbigdata.springeropen.com/articles/10.1186/s40537-019-0197-0spa
dc.relation.referencesA. Mikolajczyk and M. Grochowski, “Data augmentation for improving deep learning in image classification problem,” in 2018 International Interdisciplinary PhD Workshop (IIPhDW). IEEE, 2018, pp. 117–122. [Online]. Available: https://ieeexplore.ieee.org/document/8388338spa
dc.relation.referencesK. el Hindi and A.-A. Mousa, “Smoothing decision boundaries to avoid overfitting in neural network training,” Neural Network World, vol. 21, no. 4, p. 311, 2011. [Online]. Available: https://www.researchgate.net/publication/272237391_Smoothing_decision_boundaries_to_avoid_overfitting_in_neural_network_trainingspa
dc.relation.referencesH. Jabbar and R. Z. Khan, “Methods to avoid over-fitting and under-fitting in supervised machine learning (comparative study),” Computer Science, Communication and Instrumentation Devices, vol. 70, 2015spa
dc.relation.referencesK.-j. Kim, “Artificial neural networks with evolutionary instance selection for financial forecasting,” Expert Systems with Applications, vol. 30, no. 3, pp. 519–526, 2006. [Online]. Available: https://doi.org/10.1016/j.eswa.2005.10.007spa
dc.relation.referencesN. Srivastava, “Improving neural networks with dropout,” Master’s thesis, University of Toronto, 2013. [Online]. Available: http://www.cs.toronto.edu/∼nitish/msc thesis.pdfspa
dc.relation.referencesN. Srivastava, G. Hinton, A. Krizhevsky, I. Sutskever, and R. Salakhutdinov, “Dropout: a simple way to prevent neural networks from overfitting,” The journal of machine learning research, vol. 15, no. 1, pp. 1929–1958, 2014. [Online]. Available: https://jmlr.org/papers/v15/srivastava14a.htmlspa
dc.relation.referencesJ. Ba and B. Frey, “Adaptive dropout for training deep neural networks,” Advances in neural information processing systems, vol. 26, 2013. [Online]. Available: https://proceedings.neurips.cc/paper/2013/file/7b5b23f4aadf9513306bcd59afb6e4c9-Paper.pdfspa
dc.relation.referencesB. Ko, H.-G. Kim, K.-J. Oh, and H.-J. Choi, “Controlled dropout: A different approach to using dropout on deep neural network,” in 2017 IEEE International Conference on Big Data and Smart Computing (BigComp). IEEE, 2017, pp. 358–362. [Online]. Available: https://ieeexplore.ieee.org/document/7881693spa
dc.relation.referencesD. Molchanov, A. Ashukha, and D. Vetrov, “Variational dropout sparsifies deep neural networks,” in International Conference on Machine Learning. PMLR, 2017, pp. 2498–2507. [Online]. Available: https://arxiv.org/abs/1701.05369spa
dc.relation.referencesG. Zhang, C. Wang, B. Xu, and R. Grosse, “Three mechanisms of weight decay regularization,” in International Conference on Learning Representations, 2018. [Online]. Available: https://www.researchgate.net/publication/328598833_Three_Mechanisms_of_Weight_Decay_Regularizationspa
dc.relation.referencesS. J. Nowlan and G. E. Hinton, “Simplifying neural networks by soft weight sharing,” in The Mathematics of Generalization. CRC Press, 2018, pp. 373–394. [Online]. Available: https://ieeexplore.ieee.org/document/6796174spa
dc.relation.referencesR. Ghosh and M. Motani, “Network-to-network regularization: Enforcing occam’s razor to improve generalization,” Advances in Neural Information Processing Systems, vol. 34, pp. 6341–6352, 2021. [Online]. Available: https://proceedings.neurips.cc/paper/2021/file/321cf86b4c9f5ddd04881a44067c2a5a-Paper.pdfspa
dc.relation.referencesB. Neal, S. Mittal, A. Baratin, V. Tantia, M. Scicluna, S. Lacoste-Julien, and I. Mitliagkas, “A modern take on the bias-variance tradeoff in neural networks,” arXiv preprint arXiv:1810.08591, 2018. [Online]. Available: https://arxiv.org/abs/1810.08591spa
dc.relation.referencesP. Nakkiran, G. Kaplun, Y. Bansal, T. Yang, B. Barak, and I. Sutskever, “Deep double descent: Where bigger models and more data hurt,” Journal of Statistical Mechanics: Theory and Experiment, vol. 2021, no. 12, p. 124003, 2021. [Online]. Available: https://iopscience.iop.org/article/10.1088/1742-5468/ac3a74/metaspa
dc.relation.referencesZ. Yang, Y. Yu, C. You, J. Steinhardt, and Y. Ma, “Rethinking bias-variance trade-off for generalization of neural networks,” in International Conference on Machine Learning. PMLR, 2020, pp. 10 767–10 777. [Online]. Available: http://proceedings.mlr.press/v119/yang20j/yang20j.pdfspa
dc.relation.referencesY. Dar, V. Muthukumar, and R. G. Baraniuk, “A farewell to the bias-variance tradeoff? an overview of the theory of overparameterized machine learning,” arXiv preprint arXiv:2109.02355, 2021. [Online]. Available: https://arxiv.org/abs/2109.02355spa
dc.relation.referencesB. Ghojogh and M. Crowley, “The theory behind overfitting, cross validation, regularization, bagging, and boosting: tutorial,” arXiv preprint arXiv:1905.12787, 2019. [Online]. Available: https://arxiv.org/abs/1905.12787spa
dc.relation.referencesY. Yoshida and T. Miyato, “Spectral norm regularization for improving the generalizability of deep learning,” arXiv preprint arXiv:1705.10941, 2017. [Online]. Available: https://arxiv.org/abs/1705.10941spa
dc.relation.referencesY. Tsuzuku, I. Sato, and M. Sugiyama, “Lipschitz-margin training: Scalable certification of perturbation invariance for deep neural networks,” Advances in neural information processing systems, vol. 31, 2018. [Online]. Available: https://proceedings.neurips.cc/paper/2018/file/485843481a7edacbfce101ecb1e4d2a8-Paper.pdfspa
dc.relation.referencesH. Li, J. Li, X. Guan, B. Liang, Y. Lai, and X. Luo, “Research on overfitting of deep learning,” in 2019 15th International Conference on Computational Intelligence and Security (CIS). IEEE, 2019, pp. 78–81. [Online]. Available: https://ieeexplore.ieee.org/document/9023664spa
dc.relation.referencesA. Gavrilov, A. Jordache, M. Vasdani, and J. Deng, “Convolutional neural networks: Estimating relations in the ising model on overfitting,” in 2018 IEEE 17th International Conference on Cognitive Informatics & Cognitive Computing (ICCI* CC). IEEE, 2018, pp. 154–158. [Online]. Available: https://ieeexplore.ieee.org/document/8482067spa
dc.relation.referencesL. Deng, “The mnist database of handwritten digit images for machine learning research,” IEEE Signal Processing Magazine, vol. 29, no. 6, pp. 141–142, 2012. [Online]. Available: https://ieeexplore.ieee.org/document/6296535spa
dc.relation.referencesH. Xiao, K. Rasul, and R. Vollgraf, “Fashion-mnist: a novel image dataset for benchmarking machine learning algorithms,” arXiv preprint arXiv:1708.07747, 2017. [Online]. Available: https://arxiv.org/abs/1708.07747spa
dc.relation.referencesF. Hutter, H. H. Hoos, and K. Leyton-Brown, “Sequential model-based optimization for general algorithm configuration,” in International conference on learning and intelligent optimization. Springer, 2011, pp. 507–523. [Online]. Available: https://ml.informatik.uni-freiburg.de/wp-content/uploads/papers/11-LION5-SMAC.pdfspa
dc.relation.referencesM. López-Ibáñez, J. Dubois-Lacoste, L. Pérez Cáceres, M. Birattari, and T. Stutzle, “The irace package: Iterated racing for automatic algorithm configuration,” Operations Research Perspectives, vol. 3, pp. 43–58, 2016. [Online]. Available: https://www.sciencedirect.com/science/article/pii/S2214716015300270spa
dc.relation.referencesR. A. Fisher, “The use of multiple measurements in taxonomic problems,” Annals of eugenics, vol. 7, no. 2, pp. 179–188, 1936spa
dc.relation.referencesD. Dua and C. Graff, “UCI machine learning repository,” 2017. [Online]. Available: http://archive.ics.uci.edu/mlspa
dc.relation.referencesI. J. Goodfellow, J. Shlens, and C. Szegedy, “Explaining and harnessing adversarial examples,” in International Conference on Learning Representations, 2015. [Online]. Available: https://arxiv.org/abs/1412.6572spa
dc.relation.referencesA. Kurakin, I. J. Goodfellow, and S. Bengio, “Adversarial examples in the physical world,” in Artificial intelligence safety and security. Chapman and Hall/CRC, 2018, pp. 99–112. [Online]. Available: https://openreview.net/pdf?id=S1OufnIlxspa
dc.relation.referencesZ. Chen, Q. Li, and Z. Zhang, “Towards robust neural networks via close-loop control,” arXiv preprint arXiv:2102.01862, 2021. [Online]. Available: https://arxiv.org/abs/2102.01862spa
dc.relation.referencesL. Bottcher, N. Antulov-Fantulin, and T. Asikis, “AI Pontryagin or how artificial neural networks learn to control dynamical systems,” Nature Communications, vol. 13, no. 1, pp. 1–9, 2022. [Online]. Available: https://www.nature.com/articles/s41467-021-27590-0spa
dc.relation.referencesJ. Zhuang, N. C. Dvornek, S. Tatikonda, and J. S. Duncan, “MALI: A memory efficient and reverse accurate integrator for neural ODEs,” arXiv preprint arXiv:2102.04668, 2021. [Online]. Available: https://arxiv.org/abs/2102.04668spa
dc.relation.referencesC. Rackauckas, M. Innes, Y. Ma, J. Bettencourt, L. White, and V. Dixit, “Diffeqflux.jl-a julia library for neural differential equations,” arXiv preprint arXiv:1902.02376, 2019. [Online]. Available: https://arxiv.org/abs/1902.02376spa
dc.relation.referencesD. M. Grobman, “Homeomorphism of systems of differential equations,” Doklady Akademii Nauk SSSR, vol. 128, no. 5, pp. 880–881, 1959.spa
dc.relation.referencesP. Hartman, “A lemma in the theory of structural stability of differential equations,” Proceedings of the American Mathematical Society, vol. 11, no. 4, pp. 610–620, 1960. [Online]. Available: https://www.ams.org/journals/proc/1960-011-04/S0002-9939-1960-0121542-7/S0002-9939-1960-0121542-7.pdfspa
dc.relation.referencesMinisterio de ciencia, tecnología e innovación. Minciencias, “Guía técnica para el reconocimiento de actores del SNCTeI,” 2021. [Online]. Available: https://minciencias.gov.co/sites/default/files/upload/reconocimiento/m601pr05g07_guia_tecnica_para_el_reconocimiento_del_centro_de_desarrollo_tecnologico_cdt_v00_0.pdfspa
dc.rights.accessrightsinfo:eu-repo/semantics/openAccessspa
dc.rights.licenseAtribución-CompartirIgual 4.0 Internacionalspa
dc.rights.urihttp://creativecommons.org/licenses/by-sa/4.0/spa
dc.subject.ddc510 - Matemáticas::519 - Probabilidades y matemáticas aplicadasspa
dc.subject.proposalContinuidad Lipschitzspa
dc.subject.proposalGeneralizaciónspa
dc.subject.proposalODENetspa
dc.subject.proposalRegularizaciónspa
dc.subject.proposalRedes neuronalesspa
dc.subject.proposalResNetspa
dc.subject.proposalSobreajustespa
dc.subject.proposalGeneralizationeng
dc.subject.proposalLipschitz continuityeng
dc.subject.proposalNeural Networkseng
dc.subject.proposalOverfittingeng
dc.subject.proposalRegularizationeng
dc.subject.proposalResNeteng
dc.titleEstudio de la reducción del sobreajuste en arquitecturas de redes neuronales residuales ResNet en un escenario de clasificación de patronesspa
dc.title.translatedStudy of overfitting reduction in residual neural network architectures (ResNet) in a pattern classification scenarioeng
dc.typeTrabajo de grado - Maestríaspa
dc.type.coarhttp://purl.org/coar/resource_type/c_bdccspa
dc.type.coarversionhttp://purl.org/coar/version/c_ab4af688f83e57aaspa
dc.type.contentImagespa
dc.type.contentTextspa
dc.type.driverinfo:eu-repo/semantics/masterThesisspa
dc.type.versioninfo:eu-repo/semantics/acceptedVersionspa
dcterms.audience.professionaldevelopmentBibliotecariosspa
dcterms.audience.professionaldevelopmentEstudiantesspa
dcterms.audience.professionaldevelopmentInvestigadoresspa
dcterms.audience.professionaldevelopmentMaestrosspa
dcterms.audience.professionaldevelopmentPúblico generalspa
oaire.accessrightshttp://purl.org/coar/access_right/c_abf2spa

Archivos

Bloque original

Mostrando 1 - 1 de 1
Cargando...
Miniatura
Nombre:
1085325637.2023.pdf
Tamaño:
3.74 MB
Formato:
Adobe Portable Document Format
Descripción:
Tesis de Maestría en Ciencias - Matemática Aplicada

Bloque de licencias

Mostrando 1 - 1 de 1
No hay miniatura disponible
Nombre:
license.txt
Tamaño:
5.74 KB
Formato:
Item-specific license agreed upon to submission
Descripción: