Bibliographic citations
This is an automatically generated citacion. Modify it if you see fit
Bejarano, G., (2021). Machine Learning Models for Designing Smart Cities and Communities [Tesis, State University of New York at Binghamton]. https://renati.sunedu.gob.pe/handle/sunedu/3676921
Bejarano, G., Machine Learning Models for Designing Smart Cities and Communities [Tesis]. US: State University of New York at Binghamton; 2021. https://renati.sunedu.gob.pe/handle/sunedu/3676921
@phdthesis{renati/9247,
title = "Machine Learning Models for Designing Smart Cities and Communities",
author = "Bejarano Nicho, Gissella María",
publisher = "State University of New York at Binghamton",
year = "2021"
}
Title: Machine Learning Models for Designing Smart Cities and Communities
Authors(s): Bejarano Nicho, Gissella María
Advisor(s): Ramesh, Arti
Keywords: Aprendizaje automático; Ciudades inteligentes; Recursos hídricos; Desagregación de energía
OCDE field: https://purl.org/pe-repo/ocde/ford#1.02.01
Issue Date: 2021
Institution: State University of New York at Binghamton
Abstract: Descubrir patrones importantes en los datos puede ayudar a las ciudades a planificar, monitorear y
asignar recursos de manera más eficiente, convirtiéndolas en ciudades inteligentes con comunidades
más organizadas. Los modelos de aprendizaje automático pueden aprovechar esta gran cantidad de
datos para mejorar y escalar las tareas de estas ciudades. En este trabajo, exploramos enfoques de
aprendizaje automático para resolver diferentes problemas en el ámbito de las ciudades inteligentes
relacionados con el consumo de agua, el consumo de energía y los eventos de emergencia. Más
específicamente, nuestro trabajo arroja luz sobre el diseño de aprendizaje en conjunto, modelos
secuenciales y la combinación de modelos gráficos probabilísticos y de aprendizaje profundo para
este tipo de problemas. Además, comparamos, adaptamos e implementamos métodos
cuidadosamente para abordar las características particulares de los datos y los problemas de las
ciudades inteligentes.
Nos centramos en cuatro problemas específicos: i) clasificar el estado de operación, calidad y
cantidad de las bombas de agua, ii) predecir el consumo futuro de agua basado en el consumo
histórico, iii) predicción de la resolución temporal para eventos de emergencia y iv) desagregar
señales de energía en sus componentes de electrodomésticos. Para abordar estos problemas,
desarrollamos, comparamos y combinamos tres familias de modelos de aprendizaje automático:
aprendizaje en conjunto, gráficos probabilísticos y aprendizaje profundo.
Discovering important patterns in data can help cities to plan, monitor, and assign resources more efficiently, converting them in smart cities with more organized communities. Machine learning models can take advantage of this large amount of data to improve and scale these cities’ duties. In this work, we explore machine learning approaches to solve different problems in the smart cities domain related to water consumption, energy consumption and emergency events. More specifically, our work sheds light on the design of ensemble learning, sequential models and the combination of probabilistic graphical and deep learning models to this type of problems. Moreover, we carefully compare, adapt and implement methods to address the particular characteristics of the data and the problems of smart cities. We focus on four specific problems: i) classifying the water pump operation status, quality and quantity, ii) predicting the future water consumption based on historical consumption, iii) time resolution prediction for emergency events and iv) disaggregating energy signals into their component appliances. To address these problems, we develop, compare and combine three families of machine learning models: ensemble learning, probabilistic graphical and deep learning.
Discovering important patterns in data can help cities to plan, monitor, and assign resources more efficiently, converting them in smart cities with more organized communities. Machine learning models can take advantage of this large amount of data to improve and scale these cities’ duties. In this work, we explore machine learning approaches to solve different problems in the smart cities domain related to water consumption, energy consumption and emergency events. More specifically, our work sheds light on the design of ensemble learning, sequential models and the combination of probabilistic graphical and deep learning models to this type of problems. Moreover, we carefully compare, adapt and implement methods to address the particular characteristics of the data and the problems of smart cities. We focus on four specific problems: i) classifying the water pump operation status, quality and quantity, ii) predicting the future water consumption based on historical consumption, iii) time resolution prediction for emergency events and iv) disaggregating energy signals into their component appliances. To address these problems, we develop, compare and combine three families of machine learning models: ensemble learning, probabilistic graphical and deep learning.
Link to repository: https://renati.sunedu.gob.pe/handle/sunedu/3676921
Discipline: Ciencias de la Computación
Grade or title grantor: State University of New York at Binghamton. Thomas J. Watson College of Engineering and Applied Science
Grade or title: Doctor en Filosofía - Ciencias de la Computación
Juror: Ramesh, Arti; Seetharam, Anand; Zhang, Zhongfei (Mark); Yu, Lei; Qiao, Xingye
Register date: 19-Jun-2024
Files in This Item:
File | Description | Size | Format | |
---|---|---|---|---|
BejaranoNichoGM.pdf | Tesis | 1.54 MB | Adobe PDF | View/Open |
Autorizacion.pdf Restricted Access | Autorización del registro | 238.34 kB | Adobe PDF | View/Open Request a copy |
This item is licensed under a Creative Commons License