IEEE Symposium Series on Computational Intelligence 2009

Tutorial MCDM 1: Metaheuristics for Multiobjective Optimization

Tuesday, March 31, 8:30AM-10:30AM, Room: Tulip Grove E

Presenter: Carlos A. Coello Coello
CINVESTAV-IPN
Mexico

This tutorial provides with a general picture of the current state-of-the-art in multiobjective optimization using metaheuristics. First, some historical background is provided, dating back to the origins of multiobjective optimization in general. This discussion motivates the use of metaheuristics for solving multiobjective problems and includes a brief description of some of the earliest approaches proposed in the literature.

Then, a discussion on different heuristics used for multiobjective optimization is provided. This discussion includes evolutionary algorithms, simulated annealing, tabu search, scatter search, the ant system, particle swarm optimization and artificial immune systems.

The tutorial finishes with a discussion of some of the research topics that seem more promising for the next few years.

Tutorial MCDM 2: Evolutionary Programming with Diversity Enhancement and Ensemble Strategies

Tuesday, March 31, 4:30PM-6:30PM, Room: Tulip Grove E

Presenter: Ponnuthurai Nagaratnam Suganthan
Nanyang Technological University
Singapore

The first part of the tutorials will cover various aspects of the evolutionary programming for real parameter optimization. We will introduce various mutation operators and various strategy parameter adaptation methods including the recently proposed adaptive evolutionary programming. We will also discuss hybridization of the evolutionary programming with other optimization paradigms. Over the last 4-5 decades, researchers have proposed several approaches and alternatives to construct evolutionary algorithms. Some such alternatives are Gaussian, Levy, and Cauchy mutation operators; clearing, crowding, restricted tournament selection, and sharing based niching algorithms; adaptive penalty, epsilon, superiority of feasible, and stochastic ranking constraint handling approaches; and so on. Clearly, there are several alternative approaches at every stage of constructing an evolutionary algorithm and users will have to perform numerous simulations to pick the best approaches and to fine tune parameters. This selection and subsequent parameter tuning approach is not efficient.

In this tutorial, we will present an ensemble strategy with evolutionary programming which benefits from the need to tune parameters and the availability of diverse approaches. Our research has shown the general applicability of the ensemble strategy in solving diverse problems.

A common problem faced by almost all evolutionary optimization algorithms is premature convergence when solving hard optimization problems. In order to tackle this problem, we describe our recently proposed strategies to enhance diversity while enhancing exploitation as well. We will also present a multi-objective evolutionary programming (MOEP) which treats each objective independently, but without using the computationally expensive non-domination sorting algorithm. This MOEP is up to 20 times faster compared to the same MOEP implementation with non-domination sorting algorithm.

Tutorial MCDM 3:An Introduction to Computational Intelligence in Multi-Criteria Decision-Making: The Intersection of Search, Preference Tradeoff, and Interaction Visualization

Wednesday, April 1, 8:30AM-10:30AM, Room: Tulip Grove E

Presenter: Piero P. Bonissone
GE Global Research
USA

We consider Multi Criteria Decision Making (MCDM) as the conjunction of three components: search, preference tradeoffs, and interactive visualization. The first MCDM component is the search process over the space of possible solutions to identify the non-dominated solutions that compose the Pareto set. The second component is the preference tradeoff process to select a single solution (or a small subset of solutions) of the Pareto set. The third component is the interactive visualization process to embed the decision-maker in the solution refinement and selection loop. We focus on the intersection of these three components and we highlight some research challenges, representing gaps in the intersection. We introduce a requirement framework to compare most MCDM problems, their solutions, and analyze their performances.