|
|
 |
 |
 |
 |
 |
 |
 |
| Title | Things you may not know about evolutionary computation |
| Speaker | A/Prof. Hussein Abbass, ITEE, UNSW@ADFA |
| Date | Wednesday 29 March |
| Time | Main Seminar 11:00 - 12:00 and Tutorial 9:00-11:00 |
| Venue | Building 15 - Rm 152 |
| Abstract |
Have you used Evolutionary computation (EC) and it did not work? Have you used EC and wondered how to tune the parameters and what is going on when it works once and does not work many times? Do you want to hear about EC that works? If you answer yes to any of these questions, you should attend this seminar.
Evolutionary computation (EC) has been a controversial topic for a while. This controversy exists because of (1) a misunderstanding because researchers do not differentiate between the use of (a) EC as a problem solving technique; (b) EC as a computational framework to understand biological phenomena; and (c) EC to understand EC (i.e. develop the theory). (2) a misuse of EC techniques either because of the previous three sources of misunderstanding, or because of assuming that EC is the magic that will solve all problems. In this seminar, I will offer a gentle introduction to evolutionary computation as a problem solving technique; thus I will focus on evolutionary computation techniques that work, leaving out the hocus pocus of biologically motivated concepts. The seminar will include a few concepts from probability theory. These concepts will not be explained in the seminar not to confuse the public audience. However, I will run a 2-hour tutorial before the seminar introducing probabilities, Markov Chain Monte Carlo sampling techniques, and Bayesian inference including Bayesian networks. During these 2 hours before the seminar, I will assume that the audience are familiar with basic statistics and the tutorial will be given on the level of a first year post-graduate course. |
