Nature-Inspired Metaheuristic Algorithms for Global Optimization and Computational Intelligence

TUTORIAL:

Nature-Inspired Metaheuristic Algorithms for Global Optimization and Computational Intelligence

Presenter: Xin-She Yang (National Physical Laboratory, UK)

Date: Sunday, September 18, 2011 (afternoon)

 

Objective: To introduce the start-of-the-art metaheuristic algorithms and their applications in global optimization and computational intelligence.

Target Audience: MSc and PhD students, researchers new in the field of optimization and computational intelligence.

Prerequisite knowledge: Knowledge of basic calculus, and matrix algebra.

Timeliness of the Tutorials: Many problems in global optimization and computational intelligence are NP-hard, and there is no efficient algorithm to tackle such problems. In many cases, metaheuristic algorithms such as genetic algorithms (GA) and particle swarm optimization (PSO) are the only alternative. Over the last two decades, nature-inspired metaheuristic algorithms are becoming increasingly popular and promising in solving large-scale, nonlinear, global optimization with many real-world applications. New algorithms emerge almost every year, and this tutorial course will review and introduce some of the last developments.

Contents:

This ½-day course intends to introduce the fundamentals and latest advances of the state-of-the-art metaheuristic algorithms with the focus on the implementation and algorithm analysis. We will introduce and discuss in detail both the conventional algorithms and new metaheuristics.

1. Conventional algorithms (1 hour)

Introduce and review the conventional techniques, including

• gradient-based methods (Hill-climbing, Conjugate Gradient etc)
• derivative-free methods (Downhill simplex, pattern search etc)
• convex optimization (quadratic programming, interior-point methods etc)

2. Metaheuristic Algorithms (2.5 hours)

Introduce in detail state-of-the-art nature-inspired metaheuristics, including

• genetic algorithms, differential evolution
• simulated annealing,
• ant and bee algorithms
• particle swarm optimization,
• firefly algorithms, cuckoo search,
• harmony search and others.

3. Algorithm Analysis (1 hour)

Provide mathematical analysis of metaheuristic algorithms, including

• Algorithm convergence and complexity
• Random walks, Markov chain and Markov chain Monte Carlo
• No-free-lunch theorems and open problems

In addition, we will use a few, well-chosen case studies to demonstrate how to use each algorithm for global optimization and other intelligent applications. We also discuss the choice of the right algorithms for a given problem of interest.

This half-day short course is especially suitable for young researchers, MSc and PhD students and researchers who wish to try new techniques and new applications. We will also provide some Matlab or Octave codes for these algorithms and more detailed lecture notes will be provided as supplementary materials so that researchers can apply these methods in their own research.

Recommended References:

1. S. Boyd and L. Vandenberghe L., Convex Optimization, Cambridge University Press, (2004)
2. X. S. Yang, Nature-Inspired Metaheuristic Algorithms, Luniver Press, First Edition (2008), 2nd Edition, (2010).
3. X. S. Yang, Engineering Optimization: An Introduction with Metaheuristic Applications, John Wiley & Sons, (2010)
4. F. Glover and G. A. Kochenberger, Handbook of Metaheuristics, Kluwer Academic Publishers, (2003).
5. S. Koziel and X.-S. Yang, Computational Optimization, Methods and Algorithms, Springer, (2011).
6. K. Price, R. Storn and J. Lampinen, Differential Evolution: A Practical Approach to Global Optimization, Springer, (2005).
7. A. P. Engelbrecht, Computational Intelligence: An Introduction, 2^nd Edition, Wiley & Sons, (2007).