Implementation of new optimization algorithms in the pyMOS library

Botezatu, Mihai (2022). Implementation of new optimization algorithms in the pyMOS library. Proyecto Fin de Carrera / Trabajo Fin de Grado, E.T.S. de Ingenieros Informáticos (UPM), Madrid, España.

Description

Title: Implementation of new optimization algorithms in the pyMOS library
Author/s:
  • Botezatu, Mihai
Contributor/s:
Item Type: Final Project
Degree: Grado en Ingeniería Informática
Date: March 2022
Subjects:
Faculty: E.T.S. de Ingenieros Informáticos (UPM)
Department: Arquitectura y Tecnología de Sistemas Informáticos
Creative Commons Licenses: Recognition - No derivative works - Non commercial

Full text

[thumbnail of TFG_MIHAI_BOTEZATU.pdf] PDF - Requires a PDF viewer, such as GSview, Xpdf or Adobe Acrobat Reader
Download (2MB)

Abstract

Esta sección describe el trabajo llevado a cabo durante los últimos meses de cara a cumplir los objetivos del proyecto. Las primeras semanas estuvieron dedicadas a explorar y conocer la librería para entender su funcionamiento y uso de cara a posteriormente poder trabajar sobre ella. Así mismo, llevé a cabo una investigación general respecto al estado del arte en la materia de trabajo. Posteriormente, pasé a implementar el algoritmo CMA-ES para ser integrado en el framework. Tras la implementación de CMA-ES llevé a cabo una actualización del código de la librería para que fuese capaz de actualizar los parámetros internos en ejecuciones con algoritmos híbridos. En una siguiente fase, se trató la necesidad de implementar el algoritmo Harmony Search por lo que la llevé a cabo para ser incluida en la librería junto al algoritmo previamente implementado CMA-ES. Este segundo algoritmo, a su vez, incluye cuatro variantes distintas implementadas modularmente. Finalmente, se amplió el desarrollo implementando Parallel Chaotic Local Search Enhanced Harmony Search (PCL-MHS) como mejora del algoritmo anterior. Para validar estos desarrollos se ha llevado a cabo un benchmark comparativo de los algoritmos del que se ha extraído información de su desempeño en múltiples ejecuciones con funciones diversas.---ABSTRACT---This section describes the work carried out during the last months towards completing the objectives of this project. The first weeks were dedicated to an exploration and study of the library to understand its structure and functioning towards being able to work with it in the next phases of the project. At the same time, some research was made in relation with the state of the art in the area of study of this project. The next weeks were focused on developing the CMA-ES algorithm to be included in the framework. Afterwards, an update was made on the library to add the capability of updating internal parameters when using hybrid algorithms. In the next phase of the project, the necessity of implementing the Harmony Search algorithm was treated. Therefore, Harmony Search was implemented and added to the library together with the previous algorithm, CMA-ES. The Harmony Search algorithm included four variants which were modularly included. Finally, the development was extended by implementing Parallel Chaotic Local Search Enhanced Harmony Search (PCL-MHS) as an improvement to the previous algorithm. In order to validate the newly added implementations, a benchmarking process was carried out, extracting performance data of the algorithms in multiple executions with different functions.

More information

Item ID: 71431
DC Identifier: https://oa.upm.es/71431/
OAI Identifier: oai:oa.upm.es:71431
Deposited by: Biblioteca Facultad de Informatica
Deposited on: 28 Jul 2022 12:02
Last Modified: 28 Jul 2022 12:02
  • Logo InvestigaM (UPM)
  • Logo GEOUP4
  • Logo Open Access
  • Open Access
  • Logo Sherpa/Romeo
    Check whether the anglo-saxon journal in which you have published an article allows you to also publish it under open access.
  • Logo Dulcinea
    Check whether the spanish journal in which you have published an article allows you to also publish it under open access.
  • Logo de Recolecta
  • Logo del Observatorio I+D+i UPM
  • Logo de OpenCourseWare UPM