If a Computer Program had 2.6 Children, would it be Average?

Professor David C. Brown
Department of Computer Science, WPI

September 6, 2002
11 a.m. - 12 noon
Fuller Labs 320

Abstract

Genetic Algorithms (GAs) are one of the most versatile tools available for computer scientists interested in synthesis. They are used for Optimization, as a powerful search technique for Design and Configuration problems, and as a model of Creativity. Of interest to us is that they have been used to write programs (i.e., Genetic Programming). A GA works by operating on a "population" of possible solutions to the problem at hand. It creates a new, and hopefully improved, population by doing mutations and crossovers on selected members of the population. A mutation is a random small change, while a crossover combines selected features of two members of the population to create two new members. The members of the new population are evaluated for their "fitness" as a solution to the problem. The whole process repeats using just the fitter members. This whole process is modelled on human evolution. If the starting population is a set of simple programs, then the GA process can evolve larger, increasingly fitter programs that can solve a particular given problem. In this talk we will review this and other applications of evolutionary approaches to synthesis.

Refreshments will be served in FL 320 beginning at 10:50 a.m.

Maintained by webmaster@wpi.edu
Last modified: Sep 27, 2006, 16:05 EDT