Tutorials

1. Particle Swarm Optimization: A Universal Optimizer?

Andries Engelbrecht
University of Pretoria, Pretoria, South Africa
Email: [email protected]
Website: http://cirg.cs.up.ac.za


Abstract:
The main objective of this tutorial will be to answer the question if particle swarm optimization (PSO) can be considered as a universal optimizer. In the context of this tutorial, this means that the PSO can be applied to a wide range of optimization problem types as well as search domain types. The tutorial will start with a very compact overview of the original, basic PSO. Some experience and background on PSO will be assumed. A summary of important theoretical findings about PSO, in particular particle trajectories and convergence behavior will be provided, as this will provide important insights to the remainder of the tutorial. This will be followed by a short discussion on heuristics to select proper values for control parameters. The remainder and bulk of the tutorial will cover a classification of different problem types, and will show how PSO can be applied to solve problems of these types. This part of the tutorial will be organized in the following sections, one for each problem type:
· Continuous-valued versus discrete-valued domains
· Unimodal versus multi-modal landscapes
· Multi-solution problems requiring niching capabilities
· Constrained versus unconstrained problems, also covering boundary constraints
· Multi-objective optimization
· Dynamic environments
· Dynamic Multi-objective optimization
· Optimization with dynamically changing constraints

For each problem type, it will be shown why the standard PSO can not solve these types of problems efficiently. Simple adaptations to the PSO that will allow it to solve each problem type will then be discussed. The focus will be on PSO adaptations that do not violate the foundational principles of PSO. For each of these problem types a small subset of the most successful algorithms will be discussed.

Short Bio of Presenter:
Andries Engelbrecht is a Full Professor in Computer Science at the Department of Computer Science, University of Pretoria,and South African Research Chair in AI. He manages a research group of 40 Masters and PhD students, most of whom do research in swarm intelligence. He has recently authored a book, Fundamentals of Computational Swarm Intelligence, published by Wiley. He is also the author of a book, Computational Intelligence: An Introduction, also published by Wiley. He has presented tutorials on PSO and Co-evolutionary methods for evolving game agents at IEEE CEC 2005 and IEEE CIG 2005, respectively. He is co- presenter of a tutorial on PSO and DE at IEEE CEC 2007, and PSO at GECCO 2007. He also presented PSO tutorials at ACISS and ACAL 2010, CEC 2009, CEC 2012, CEC 2013, GECCO 2013, and to a number of universities. He has published approximately 200 papers in the last decade, serves as a reviewer for a number of conferences and journals, and is an associate-editor of IEEE TEC, IEEE TCIAIG, and Swarm Intelligence, and serves on the editorial board of three other journals. He served as a member of a large number of conference program committees, and is in the organizing committee of several conferences.


 

2. Introduction to Dynamic Multi-objective Optimization and its Challenges

Mardé Helbig
University of Pretoria, South Africa
Email: [email protected]
Website: http://up-za.academia.edu/MardeHelbig

 

Abstract:
Most optimization problems in real-life have more than one objective, with at least two objectives in conflict with one another and at least one objective that changes over time. These kinds of optimization problems are referred to as dynamic multi-objective optimization (DMOO) problems.

Instead of re-starting the optimization process after a change in the environment has occurred, previous knowledge is used and if the changes are small enough, this may lead to new solutions being found much quicker.

Most research in multi-objective optimization has been conducted on static problems and most research on dynamic problems has been conducted on single-objective optimization. The goal of a DMOO algorithm (DMOA) is to find an optimal set of solutions that is as close as possible to the true set of solutions (similar to static MOO) and that contains a diverse set of solutions. However, in addition to these goals a DMOA also has to track the changing set of optimal solutions over time. Therefore, the DMOA also has to deal with the problems of a lack of diversity and outdated memory (similar to dynamic single-objective optimization).

This tutorial will introduce the participants to the field of DMOO by discussing: benchmark functions and performance measures that have been proposed and the issues related to each of these; algorithms that have been proposed to solve DMOO; issues with comparison of DMOAs performance and ensuring a fair comparison; analysing the results and why traditional approaches used for static MOO is not necessarily adequate enough; challenges in the field that provide interesting research opportunities.

Short Bio of Presenter:
Mardé Helbig is a senior lecturer at the University of Pretoria, South Africa. She obtained her PhD in 2012 at the University of Pretoria, with a thesis entitled: “Solving dynamic optimisation problems using the vector evaluated particle swarm optimisation algorithm”. She has been the main organizer of special sessions on DMOO at CEC 2014, CEC 2015 and a competition on DMOO at CEC 2015. She also presented a tutorial on DMOO at SSCI 2015. She has numerous publications on DMOO and is a regular reviewer for the top conferences and journals in the field. In addition, she is the vice chair of the IEEE Task force on Evolutionary Multi-objective Optimization, the chair of the IEEE Computational Intelligence Chapter in South Africa and a member of the IEEE Woman in Computational Intelligence sub-committee.



3. CIlib: A Computational Intelligence Library


Gary Pampara and Andries Engelbrecht
University of Pretoria, Pretoria, South Africa
Tel.& Fax: +27 12 420 3578 (phone), +27 12 420 5188 (fax),
Email: [email protected], [email protected],


Abstract
This tutorial will provide a short introduction to a software library designed specifically to aid in CI research. This talk will rstly show that the current mechanisms and general workflow for CI research, has many pitfalls and misunderstandings that are not normally catered for. A discussion how these concerns are addressed by the software library will be shown and how these concerns are handled in a principled manner. The talk will also, briefly, introduce how functional programming can benefit the CI community by exploiting common abstractions and derived operations. The software library will be discussed, with focus on the core design decisions that have been made, for the sole purpose of aiding CI research. From these designs, several CI speci c abstractions have been identified, and formalized, that model CI algorithmic components, which facilitate the research process through allowing these individual pieces to be composed together thereby creating larger computational structures, which ultimately represent an algorithm definition. Examples of algorithmic formulation will be shown, and how such formulations may be combined to create new formulations, which range from swarm intelligence algorithms (PSO, DE, GA, etc) to other algorithmic formulations such as cooperative evolution, hyper heuristics and multi-objective computation. Experimentation will be demonstrated, whereby algorithmic components are composed together in an experimental environment, removing the need to explicitly write programs to test ideas. As a result of this tutorial, researchers and practitioners will gain a better understanding of these very important aspects of principled experimental design, addressing the issues highlighted and receiving guidelines of how an open source software library will aid algorithmic veri cation and experimentation.

 

Short Bio of Presenter:

Andries Engelbrecht is a Full Professor in Computer Science at the Department of Computer Science, University of Pretoria,and South African Research Chair in AI. He manages a research group of 40 Masters and PhD students, most of whom do research in swarm intelligence. He has recently authored a book, Fundamentals of Computational Swarm Intelligence, published by Wiley. He is also the author of a book, Computational Intelligence: An Introduction, also published by Wiley. He has presented tutorials on PSO and Co-evolutionary methods for evolving game agents at IEEE CEC 2005 and IEEE CIG 2005, respectively. He is co- presenter of a tutorial on PSO and DE at IEEE CEC 2007, and PSO at GECCO 2007. He also presented PSO tutorials at ACISS and ACAL 2010, CEC 2009, CEC 2012, CEC 2013, GECCO 2013, and to a number of universities. He has published approximately 200 papers in the last decade, serves as a reviewer for a number of conferences and journals, and is an associate-editor of IEEE TEC, IEEE TCIAIG, and Swarm Intelligence, and serves on the editorial board of three other journals. He served as a member of a large number of conference program committees, and is in the organizing committee of several conferences.


Gary Pampara is a PhD candidate of the Computer Science department of the University of Pretoria, under the supervision of Professor Andries Engelbrecht. He is currently researching constraint handling and enforcement in evolutionary computation, with specific focus on PSO within the CIRG research group within the Department of Computer Science at the University of Pretoria. In order to perform the experimental work an additional research output was created, that being a tool for computational intelligence. This tool is in the form of a software library called CIlib, for which he is the lead author.


 

4. Multi-Method Algorithms: Past, Present and Future

Jacomine Grobler
Department of Industrial and Systems Engineering, University of Pretoria, South Africa,
Email: [email protected]

Abstract:

Over the last five decades a large number of optimization algorithms have been developed to address numerous real world optimization problems. Unfortunately, it is not always easy, or even possible, to predict which one of the many algorithms already in existence will be the most suitable for solving a specific problem. This unpredictability is not only limited to different algorithms on different problem classes, but there may even be issues with respect to large variations in algorithm performance over different instances of the same problem. Within this context, the simultaneous use of more than one algorithm for solving optimization problems becomes an attractive alternative.

A multi-method algorithm can be described as consisting of one or more entities representing candidate solutions which are evolved over time, a set of available algorithms or operators, referred to as constituent algorithms or operators, and a high level strategy responsible for allocating the entities to the most suitable algorithms at different stages of the optimization process.

Multi-method algorithms have appeared in various different domains over the last couple of years. Examples include memetic computation, algorithm portfolios, algorithm ensembles, heterogeneous algorithms and hyper-heuristics. This tutorial will provide an overview of each of these domains as well as an introduction to the state-of-theart algorithms of these domains. Various multi-method algorithm design considerations will also be discussed. These considerations include the selection of constituent algorithms or operators, the exchange of knowledge between constituent algorithms throughout the optimization process and the allocation of entities to constituent algorithms. Finally, unresolved challenges in the field as well as opportunities for future research, will be discussed.


Short Bio of Presenter:

Jacomine is an Industrial Engineer and senior lecturer in the Department of Industrial and Systems Engineering at the University of Pretoria in South Africa. Her main fields of expertise are multi-method optimization algorithms, swarm intelligence, multi-objective optimization, scheduling, and supply chain optimization. She completed her PhD in multi-method algorithms through the Computational Intelligence Research Group at the University of Pretoria and was recently awarded the 2015 JD Roberts emerging researcher award for her contribution to the development of mathematical models and optimization algorithms. She regularly reviews papers for leading international journals such as IEEE Transactions on Evolutionary Computation, Swarm Intelligence and Journal of Scheduling, acted as the Pretoria Chapter Chair of the Operations Research Society of South Africa in 2009, was the Registration Co-Chair for the 2015 IEEE Symposium Series in Computational Intelligence and is a current member of the IEEE Woman in Computational Intelligence sub-committee. Finally, Jacomine has already presented various invited lectures to, for example, the Women in Computational Intelligence lunch at the Symposium Series on Computational Intelligence in Orlando, USA (12/2014) and the Workshop on Nature-inspired Techniques in Computer Networking at the University of Cyprus (11/2010) and participated as a panellist at the “40th Anniversary Discussion” at the 2009 Conference of the Operations Research Society of South Africa.