http://www.dspace.espol.edu.ec/handle/123456789/53666 Doctorado en Ciencias Computacionales Aplicadas Thu, 16 Apr 2026 08:53:20 GMT 2026-04-16T08:53:20Z http://www.dspace.espol.edu.ec/handle/123456789/65802 Title: Estimating intraurban socioeconomic status using users’interactions registered on digital data Authors: Cruz Ramírez, Eduardo Segundo; Vaca Ruiz, Carmen, Dircetor Abstract: The thesis addresses the challenge of estimating socioeconomic status (SES) at an intraurban level using digital data sources. Traditional methods for measuring SES, such as censuses and surveys, are often limited by their infrequency and coarse spatial granularity, which hinders timely and accurate assessments, especially at the neighborhood level. The study proposes leveraging alternative digital data sources, including mobile phone top-up transactions and supermarket purchase data, to model and predict SES, providing the potential for more frequent, cost-effective, and spatially granular analysis. The research focuses on urban neighborhoods in Ecuador, aiming to develop machine learning models that can accurately predict Neighborhood SES (NSES). The research employs two machine learning models: a Regression Model using mobile phone top-up transactions and a Graph Neural Network (GNN) Model using supermarket transaction data. The first model focuses on linear relationships between variables derived from top-up transaction data and NSES. The model is designed to estimate the NSES by aggregating the average denomination and the denomination diversity at the neighborhood level. The second model leverages the complex, non-linear relationships inherent in supermarket transactions. The GNN model transforms these transactions into a graph representation, where items purchased together are linked, and the frequency and diversity of these links are analyzed to infer SES. The model is particularly suited for capturing the socioeconomic patterns that emerge from the co-purchase behaviors of individuals within a neighborhood. Both models demonstrate significant predictive power in estimating SES at the intraurban level. The Regression Model achieves a prediction accuracy of up to 74%. This model is particularly effective in identifying the relationship between average top-up denomination and neighborhood SES, with higher denominations indicating wealthier neighborhoods. The GNN Model outperforms the Regression Model, achieving a prediction accuracy of up to 91%. The GNN model is able to model the intricate patterns of co-purchases within neighborhoods, allowing for a more detailed and accurate representation of NSES. The results highlight the potential of digital data sources as viable alternatives to complement traditional SES measurement methods. Description: x Fri, 11 Apr 2025 00:00:00 GMT http://www.dspace.espol.edu.ec/handle/123456789/65802 2025-04-11T00:00:00Z http://www.dspace.espol.edu.ec/handle/123456789/57657 Title: Human pose estimation based in Deep Learning techniques from Multi-view Environments Authors: Vintimilla, Boris X., Director; Sappa, Angel D., Co-Director Description: Esta tesis presenta una serie de técnicas basadas en CNN aplicadas a imágenes para abordar los problemas de estimación de la pose de la cámara y del cuerpo humano a partir de entornos multivistas. Para la estimación de la pose de la cámara, se han propuesto dos enfoques basados en la arquitectura siamesa para estimar los parámetros extrínsecos de la pose de la cámara. El primer enfoque toma como entrada un conjunto de pares de imágenes reales, que deben tener un solapamiento mínimo para asegurar que los pares de imágenes tienen características comunes. Sin embargo, debido a los pocos conjuntos de datos de imágenes reales disponibles para la estimación de la pose de la cámara en escenarios multivista, se propone un segundo enfoque. Este consiste en una estrategia de adaptación del dominio, que incluye la generación de diferentes escenarios virtuales mediante un software especial desimulación3D.La estrategia se utiliza para aprovechar la transferencia del conocimiento aprendido de estos escenarios virtuales a los escenarios del mundo real. Para el problema de estimación de la postura del cuerpo humano, también se proponen dos enfoques. El primero, una arquitectura basada en una red neuronal convolucional, que aprovecha los parámetros extrínsecos estimados para establecer la relación entre las diferentes cámaras en el esquema multivista. Esto ha permitido estimar la pose del cuerpo humano utilizando información de diferentes puntos de vista, y así, resolver el desafiante problema de la auto-oclusión en la estimación de la pose humana debido a la pose natural del cuerpo. También se ha propuesto un segundo enfoque para el problema de la estimación de la pose del cuerpo humano. Este utiliza módulos de atención para detectar las articulaciones del cuerpo. Sin embargo, a diferencia del primer enfoque, este nuevo enfoque no tiene encuenta los parámetros extrínsecos entre las diferentes cámaras del esquema multivista, sino que se utiliza la posición y la orientación de los huesos del cuerpo humano como información adicional para abordar el problema de la auto-oclusión de las articulaciones del cuerpo humano. La precisión de estas estimaciones es importante para evitar posibles falsas alarmas en los sistemas de análisis del comportamiento de las ciudades inteligentes, así como en aplicaciones de fisioterapia y asistencia para el desplazamiento seguro de personas mayores, entre otras. Sun, 01 Jan 2023 00:00:00 GMT http://www.dspace.espol.edu.ec/handle/123456789/57657 2023-01-01T00:00:00Z http://www.dspace.espol.edu.ec/handle/123456789/57097 Title: Thermal image super-resolution using deep learning techniques Authors: Rivadeneira Campodónico, Rafael Eduardo; Sappa, .Angel D., Director; Vintimilla, Boris X., Co-Director Abstract: In recent years, there has been an increasing demand for high-resolution images, especially in the field of security and surveillance. Super-resolution is a technique that can be used to improve the resolution of an image. Most of these techniques are based on using a single image or a set of low-resolution images from the visible spectrum, where the high-resolution image is reconstructed by using a model that considers a degradation process. Nevertheless, images from the visible spectrum are limited by the atmospheric conditions and the availability of light. While human visual perception is limited to the visual-optical spectrum, machine vision isnot.This dissertation presents the use of images from the long-wavelength infrared spectral band, namely thermal images, for the purpose of super-resolving them. Thermal images are notaffectedbyatmosphericconditions,andtheycanbeacquiredeveninlow-lightconditions. In order to obtain a high-resolution image from a set of low-resolution thermal images, deep learning techniques are used, specifically convolutional neural networks. The results show that improving the thermal images’ resolution is possible while preserving the scene’s main features. Two main paths are tackled in the present work, the single and multi-image super-resolution, where a dataset with an extensive collection of images is collected to address this purpose. One of the main properties of this thesis is to show that thermal image super-resolution can be tackled by using the proposed architectures and validating them with the acquired public dataset used in several challenges. Sun, 01 Jan 2023 00:00:00 GMT http://www.dspace.espol.edu.ec/handle/123456789/57097 2023-01-01T00:00:00Z http://www.dspace.espol.edu.ec/handle/123456789/56505 Title: Towards improved task scheduling incloud computing plataforms Authors: Boza Gaibor, Edwin Federico; Abad, Cristina, Director Abstract: With the microservices architectural model, complex systems are built as a set of small loosely-coupled independent components. Deployment of microservice-based applications takes advantageof cloud computing platforms to provide scalability, high availability, and to simplify the deploymentand operation of applications. Improving the performance of microservices running on containerizedand serverless cloud computing platforms is important for many applications with requirements likereal-time, low latency, responsive auto-scaling and to increase user engagement.This thesis aims to improve the performance of these applications through smart task schedul-ing and request routing decisions. We started our work enhancing function latency on serverlessplatforms by increasing code locality, and assessing conflicting goals for the scheduling process. Wethen studied the benefits of performance-aware scheduling decisions for containerized platforms, toimprove microservices performance. We demonstrated that our affinity scheduling approach hasimportant applicability beyond the microservices domain by increasing the use of the cache layerto reduce the access time to the data stored in a data lake. Finally, we also addressed the use ofserverless-based microservices to support the dynamic implementation of self-adaptive cloud services. Sat, 01 Jan 2022 00:00:00 GMT http://www.dspace.espol.edu.ec/handle/123456789/56505 2022-01-01T00:00:00Z