Institutional Repository of Vilnius University
Dalelių spiečių optimizavimo algoritmų taikymo daugiakriteriams uždaviniams efektyvumo tyrimas /
Abstract
dc:descriptionIn this thesis the problem of evaluating the performance of multi-objective optimization methods for non-convex problems is examined. Namely, the performance of multi-objective particle swarm optimization methods are investigated. An overview of these methods is provided in this thesis covering most methods described in literature. A novel classification system of these methods is developed. This system uses method design choices to classify them. A thorough experimental analysis of existing methods is given. Each method is tested using a wide variety of test problems. The results are further analyzed with regards to what types of problems each method solves best. An important aspect of solution quality when it comes to multi-objective problems is the uniformity of solution spread along the real Pareto frontier. Due to the inadequacies of existing performance indicators when it comes to measuring Pareto frontier approximation solution spread, two new indicators are proposed. These two indicators are designed to capture the intuitive notion of solution spread uniformity. They are discussed in comparison with existing indicators. Two new multi-objective particle swarm optimization methods are proposed in the thesis as well. These methods are based on the idea of heterogeneous swarms - swarms where several different types of particles are used at the same type. The particles share information via the same non-dominated point archive.
Degree
thesis:*- Grantor dc:publisher
- Institutional Repository of Vilnius University
- Year dc:date
- 2016
Author and committee
dc:creator, dc:contributor.*- Author dc:creator
-
- Jančauskas, Vytautas,
- Contributors dc:contributor
-
- Žilinskas, Antanas
Subjects
dc:subject × 1Rights
dc:rights- Statement dc:rights
-
- info:eu-repo/semantics/openAccess
- Language dc:language
- lit
Identifiers
dc:identifier.*- Repository record dc:identifier
- https://repository.vu.lt/VU:ELABAETD16662311&prefLang=en_US