Ciencias Exactas y Ciencias de la Salud
Permanent URI for this collectionhttps://hdl.handle.net/11285/551039
Pertenecen a esta colección Tesis y Trabajos de grado de las Maestrías correspondientes a las Escuelas de Ingeniería y Ciencias así como a Medicina y Ciencias de la Salud.
Browse
Search Results
- Tailoring metaheuristics for designing thermodynamic-optimal water based cooling devices for microelectronic thermal management applications(Instituto Tecnológico y de Estudios Superiores de Monterrey, 2023-06) Pérez Espinosa, Guillermo; TERASHIMA MARIN, HUGO; 65879; Terashima Marín, Hugo; emipsanchez; Ortiz Bayliss, José Carlos; Aviña Cervantes, Juan Gabriel; Escuela de Ingeniería y Ciencias; Campus Monterrey; Cruz Duarte, Jorge MarioHeat sinks provide a common and straightforward alternative to dealing with the Microelectronic Thermal Management (MTM) problem due to their simplicity of fabrication, low cost, and reliability of heat dissipation. The MTM problem is highly relevant in today's electronics industry, as new electronic devices' miniaturization and enhanced performance have increased their heat power generation. So, regarding the second law of thermodynamics, an optimal heat sink design can guarantee that the microelectronic components operate without jeopardizing their life span and performance. To solve this challenging problem, Metaheuristics~(MHs) have shown to be excellent alternatives due to their reliability, flexibility, and simplicity. Nevertheless, no single MH guarantees an overall outstanding performance. Thus, the motivation for this work is to open ample room for practitioners to find the proper solver to deal with a given problem without requiring extensive knowledge of heuristic-based optimization. This work studies the feasibility of implementing a strategy for Automatic Metaheuristic Design powered by a hyper-heuristic search to minimize the entropy generation rate of microchannel heat sinks and tailor population-based and metaphor-less MHs for solving the MTM. A mathematical model based on thermodynamic modeling via the Entropy Generation Minimization (EGM) criterion was used to obtain the value of the entropy generation rate of a rectangular microchannel heat sink according to their design. Four different scenarios were considered, varying the design specifications for the heat sinks and comparing our generated MH against seven well-known heuristic-based algorithms from the literature. The one-sided Wilcoxon signed ranked test was used to perform these comparisons. Statistical evidence was found to claim that our tailored MHs manage to outperform them, in most cases, at least in the tested scenarios. Additionally, we followed a methodology to infer which operators should be considered in a curated heuristic space to design the proper MH easily. We found that using this curated search space benefits the overall process, as the HH algorithm managed to tailor high-performing MHs faster and more consistently than its counterpart. Furthermore, insights were obtained on which HH parameters are more suitable for our search, as some can enhance the tailoring process when tuned correctly. Finally, we tested some of our best designs found to see how they perform when minor fluctuations appear on some variables, just as they occur in real-life implementations. All the experimentation processes also found that the search operators of evolutionary algorithms are well suited to solve this problem, as they compose several of our tailored MHs, and that the combination of High Thermal Conductive Graphite and water achieved the lower entropy generation rate values from the four combinations tested.
- Detection of suspicious attitudes on video using neuroevolved shallow and deep neural networks models(Instituto Tecnológico y de Estudios Superiores de Monterrey, 2021-11) Flores Munguía, Carlos; Terashima Marín, Hugo; puemcuervo/tolmquevedo; Oliva, Diego; Ortiz Bayliss, Jose Carlos; School of Engineering and Sciences; Campus MonterreyThe analysis of surveillance cameras is a critical task usually limited by the people involved in the video supervision devoted to such a task, their knowledge, and their judgment. Security guards protect other people from different events that can compromise their security, like robbery, extortion, fraud, vehicle theft, and more, converting them to an essential part of this type of protection system. If they are not paying attention, crimes may be overlooked. Nonetheless, different approaches have arisen to automate this task. The methods are mainly based on machine learning and benefit from developing neural networks that extract underlying information from input videos. However, despite how competent those networks have proved to be, developers must face the challenging task of defining the architecture and hyperparameters that allow the network to work adequately and optimize the use of computational resources. Furthermore, selecting the architecture and hyperparameters may significantly impact the neural networks’ performance if it is not carried out adequately. No matter the type of neural network used, shallow, dense, convolutional, 3D convolutional, or recurrent; hyperparameter selection must be performed using empirical knowledge thanks to the expertise of the designer, or even with the help of automated approaches like Random Search or Bayesian Optimization. However, such methods suffer from problems like not covering the solution space well, especially if the space is made up of large dimensions. Alternatively, the requirement to evaluate the models many times to get more information about the evaluation of the objective function, employing a diverse set of hyperparameters. This work proposes a model that generates, through a genetic algorithm, neural networks for behavior classification within videos. The application of genetic algorithms allows the exploration in the hyperparameters solution space in different directions simultaneously. Two types of neural networks are evolved as part of the thesis work: shallow and deep networks, the latter based on dense layers and 3D convolutions. Each sort of network takes distinct input data types: the evolution of people’s pose and videos’ sequences, respectively. Shallow neural networks are generated by NeuroEvolution of Augmented Topologies (NEAT), while CoDeepNEAT generates deep networks. NEAT uses a direct encoding, meaning that each node and connection in the network is directly represented in the chromosome. In contrast, CoDeepNEAT uses indirect encoding, making use of cooperative coevolution of blueprints and modules. This work trains networks and tests them using the Kranok-NV dataset, which exhibited better results than their competitors on various standard metrics.
- Two-fold Approach for Video Retrieval: Semantic Vectors to Guide Neural Network Training and Video Representation Approximation Via Language-Image Models(Instituto Tecnológico y de Estudios Superiores de Monterrey, 2021-05) Portillo Quintero, Jesús Andrés; TERASHIMA MARIN, HUGO; 65879; Terashima Marín, Hugo; tolmquevedo, emipsanchez; Ortiz Bayliss, José Carlos; Han, David; Escuela de Ciencias e Ingeniería; Campus MonterreyVideo Retrieval is a challenging task concerned with recovering relevant videos from a collection to fulfill a query. Defining relevance is an unsolved problem in Information Retrieval literature since it is prone to subjective considerations. Text-based Video Retrieval systems calculate relevance by measuring the relationship between a textual query and video metadata. This is a widely used approach, but it does not consider motion and video dynamics. On the other hand, content-based methods account for visuals in the retrieval process but can only operate with visual queries. This phenomenon poses the question of whether it is possible to create a Video Retrieval system that collects video based on visual content and works with textual queries. A method to bridge the semantic gap between video and text is presented. This approach employs a Multimodal Machine Learning model capable of mapping multiple types of infor- mation among themselves. The connection between modalities occurs in a learned video-text space, where it is possible to measure similarity between them. With a trained system like this, it is possible to retrieve the most similar videos to a query by obtaining the similarity between the vector representation of a text query and a collection of videos. The work presented in this thesis is focused on a Dual Encoder architecture to funnel video and text information through independent Neural Networks. These Neural Networks take advantage of pre-trained models for each modality, called backbones. Other authors have used word-level backbones to encode text; we claim this method restricts the descriptiveness of text. One contribution from this work is the implementation of a novel sentence-level em- bedding backbone. This method generates sentence vectors representing the holistic phrasal meaning and has the added benefit of allowing to measure the semantic similarity among sen- tences. A second research contribution is to employ similarity measurements in text in order to guide the Neural Network training. A proposed Proxy Mining loss finds the contrary of sentences and their corresponding videos to ground the video-text space training. Compared to video, the image modality has been dedicated more assets and research efforts given its ease of use. The possibility of leveraging those assets to video is considered. A third scientific contribution is to extend the image and text representation called CLIP. This model is pre-trained to produce a fixed-size representation for images and text that allows for similarity measurement. By trying several aggregation methods, it was possible to collapse the temporal dimension inherent in videos, hence approximate a video-text representation. This breakthrough resulted in state-of-the-art results on the MSR-VTT and MSVD benchmarks. This document represents a thesis project for the degree of Master in Computer Science from Instituto Tecnolo ́gico y de Estudios Superiores de Monterrey.
- Hipocorísticas mediante un enfoque neuro-evolutivo para el ordenamiento dinámico de variables en problemas de satisfacción de restricciones(Instituto Tecnológico y de Estudios Superiores de Monterrey, 2008-12-01) Fuentes Rosado, Jorge Iván; FUENTES ROSADO, JORGE IVAN; 218157; Terashima Marín, Hugo; Acevedo Mascarúa, Joaquín; Valenzuela Rendón, Manuel; Conant Pablos, Santiago E.; Tecnológico de Monterrey, Campus Monterrey; División de Graduados en Mecatrónica y Tecnologías de Información; Campus MonterreyMuchos de los problemas de las Ciencias de la Computación y de la Inteligencia Artificial (IA) pueden ser caracterizados como problemas de Satisfacción de Restricciones. Un CSP, por sus siglas en inglés, está compuesto por un conjunto de variables, un conjunto dominio de posibles valores y una serie de restricciones entre variables a ser satisfechas. Encontrar una solución para un problema de satisfacción de restricciones consiste en encontrar una asignación consistente para todo el conjunto de variables de tal manera que satisfaga todas las restricciones. Los algoritmos de solución más conocidos son Backtracking, Foward Checking y otros híbridos los cuales están basados en la teoría de algoritmos de búsqueda en árboles. La solución de estos problemas cuando son considerados difíciles por métodos tradicionales se vuelve muy lento y tedioso puesto que el espacio de búsqueda es muy grande, es por ello que se han realizado muchas investigaciones para la optimización de la forma de solucionar estos problemas. Además, existen instancias de CSPs que son particularmente difíciles de resolver, ya que requieren una elevada cantidad de verificaciones de consistencia para encontrar una solución o demostrar que no existe alguna. El ordenamiento de las variables juega un papel de suma importancia en la búsqueda de la solución. Un buen ordenamiento de variables al momento de buscar una solución representa una menor complejidad en la búsqueda. Existen dos enfoques para el ordenamiento de variables. El primer enfoque es conocido como en ordenamiento estático el cual consiste en definir el orden en la cual las variables estarían instanciadas antes de empezar a solucionar el problema. El otro enfoque es conocido como ordenamiento dinámico el cual consiste en determinar la variable a instanciar en tiempo de solución. Para esto surgieron algunas heurísticas que ayudan en este proceso, de las cuales podemos nombrar Fail First, Rho, Kappa, entre otras, pero ninguna de ellas ha demostrado ser eficiente para todas las instancias. En este enfoque la heurística elegida es aplicada en todos los niveles del árbol sin importar las modificaciones del problema. Recientemente se ha empezado a trabajar con un enfoque basado en hiperheurísticas, el cual, dada las características del estado del problema determina que heurística debe ser aplicada para elegir la variable a ser asignada, en este modelo en cada nivel del árbol la heurística aplicada puede variar respecto a la anterior. Este modelo se compone de una etapa de entrenamiento y una de prueba. La fase de entrenamiento consiste en utilizar un algoritmo genético para evolucionar una población de redes neuronales. Las redes neuronales son entrenadas por medio del algoritmo de retroprogramación del error y después evaluadas con diferentes instancias de CSPs. La red neuronal tiene como entrada el estado del problema y como salida la heurística sencilla a aplicar. La intención de utilizar diferentes instancias de CSPs durante la etapa de entrenamiento es desarrollar procesos de solución generales que puedan ser aplicados para resolver un gran número de instancias, mías que encontrar buenas soluciones para instancias específicas. La idea general es la siguiente, dado el estado del CSP P es utilizado para alimentar la red neuronal de la cual se obtendrá como salida la heurística sencilla a aplicar. Una vez aplicada la heurística el CSP P cambia a un CSP P', el cual es una versión actualizada del anterior. El procedimiento se repite hasta resolver el problema completamente. Al terminar el entrenamiento de la población se toma el mejor individuo y es utilizado para solucionar tanto los problemas utilizados para su entrenamiento como problemas nuevos. Lo anterior para hacer comparaciones con los resultados de la aplicación individual de las heurísticas simples. Los resultados demuestran que es factible la creación de una hiperheurística bajo este modelo de solución puesto que las hiperheurísticas generadas se comportaron como la mejor heurística sencilla aplicada y con esto mejor que el promedio que ellas. La mejor heurística simple puede ser diferente en cada instancia del problema.
- Metodología y marco teórico para la implementación de un sistema nervioso digital dentro de una organización(Instituto Tecnológico y de Estudios Superiores de Monterrey, 2002-12-01) Andrade Sepúlveda, Juan Gerardo; TERASHIMA MARIN, HUGO; 65879; Terashima Marín, Hugo; Marcos Marcos, María Socorro; Garrido Luna, Leonardo; Programa de Graduados en Computación, Información y Comunicaciones; Campus MonterreyEl desarrollo de la investigación aborda la implementación de un sistema nervioso digital dentro de una organización, en esta investigación se pretende presentar un marco teórico de cada uno de los conceptos involucrados en el desarrollo de la implementación del sistema nervioso digital propuesto al final de la investigación, con el cual se puedan tener las bases suficientes y el soporte de autores reconocidos en cada uno los conceptos manejados en esta tesis. Uno de los principales dentro de las organizaciones es la planeación, creación y manejo de la información de manera integral, rápida, eficiente y además distribuida. Con esta premisa hace más importante contar con el conocimiento necesario y una metodología para su aplicación y con esto poder ¡mplementar un sistema nervioso digital dentro de las organizaciones. En la investigación se integran conceptos teóricos como Visión de la organización, Planeación estratégica, las Organizaciones y sus estructuras, las tecnologías de información en la organización moderna, varios esquemas para poder Alinear las Tecnologías de Información con la Estrategia de negocios de la organización, así como entender que es la Cultura organizacional, además de tener información de las redes intranet, su estructura y uso actual, para llegar así a poder entender el concepto integral de lo que es el sistema nervioso digital y su metodología propuesta en una serie de pasos a seguir, y un esquema mental a desarrollar que será muy importante dentro de la investigación.
- Detection of Violent Behavior in Open Environments Using Pose Estimation and Neural Networks(Instituto Tecnológico y de Estudios Superiores de Monterrey) Chong Loo, Kevin Brian Kwan; TERASHIMA MARIN, HUGO; 65879; Terashima Marín, Hugo; tolmquevedo, emipsanchez; Conant Pablos, Santiago Enrique; Escuela de Ingeniería y Ciencia; Campus MonterreyPeople’s safety and security have always been an issue to attend. With the coming of techno- logical advances, part of it has been used to improve safeguards, though other aspects, without precautions, have made people even more vulnerable. People can get their sensitive data stolen or become victims of transaction fraud. These may be crimes done without physical interac- tion, but felonies with physical violence still exist. Some solutions for pedestrian safety are guards, police cars patrolling, sensors and security cameras. Nonetheless, these methods only react when the crime is happening or, even more critical, when it has already occurred, and the damage has been done. Therefore, numerous methods have been implemented using Arti- ficial Intelligence in order to solve this problem. Many approaches to detect violent behavior and action recognition rely on 3D convolutional neural networks (3D CNNs), spatial tempo- ral models, long short term memory networks, pose estimation among other implementations. However, in the current state of the art, how these approaches are used do not work perfectly and are not adapted to an uncontrolled environment. Therefore, a significant contribution from this work was the development of a new solu- tion model that is able to detect violent behavior. This approach focuses on using pedestrian detection, tracking, pose estimation and neural networks to predict pedestrian behavior in video frames. This method uses a time window frame to extract joint angles, given by the pose estimation algorithm, as features for classifying behavior. At the moment of developing this thesis project, there were not many databases with violent behavior videos. The ones that existed were low quality; cluttered were pedestrians cannot be seen clearly, and with unfixed camera angles. Consequently, another important contribution of this work was creating a new database, Kranok-NV, with a total of 3,683 normal and violent videos. This database was used to train and test the solution model. For the evaluation, a protocol was designed using 10-fold cross- validation. With the implemented solution model, accuracy of more than 98% was achieved on the Kranok-NV database. This approach surpassed the performance of state of the art methods for violence detection and action recognition in the developed database. Though this new solution model is able to detect violent and normal behavior, it can be easily extended to classify more types of behaviors. Further work requires to test this approach in emerging databases of videos and optimize specific areas of the solution model. Additionally, the contributions of this work can aid in the development of new approaches.