EEL 4818 (Lectures on Genetic Algorithms)
Fall 2006
by Ivan Garibay, University of Central Florida, September 18, 2006.
Lecture 1, September 21, 2006
- Topics Covered:
-
Genetic Algorithms (GAs): What are they?
- Darwinian Evolution, Mendelian Genetics, Turing's Genetic Search, and Holland's Genetic Algorithms.
-
Stochastic Search and Evolutionary Computation
-
GA components: problem representation, selection function, fitness function, and genetic operators.
-
GA pseudocode
- Materials:
- Deadlines:
- GA section written evaluation: October 5th, 2006.
Lecture 2, September 26, 2006
- Topics Covered:
- GAs: How do they work?
- GA Optimization: One Max Problem: our class becomes a random number generator for running a single GA loop by hand
- Tournament Selection
- One-point Crossover
- Bit-flip Mutation
- Analysis of One Max Problem's experimental results.
Lecture 3, September 28, 2006
- Topics Covered
- GAs: Why do they work? (stochastic search answer)
- Universal limits of search: No Free Lunch Theorems
- Search Spaces, Fitness Landscapes
- Random, Needle-in-a-haystack, and one-max fitness landscapes
- Search Algorithms and Stochastic Search
- Why stochastic search works? Key property
- Schemas
Lecture 4, October 3, 2006
- Topics Covered:
GAs: Why do they work? (GA schema analysis answer)
- Building Block Hypothesis
- Schema Theorem (Holland, 1975)
- How does GA process schemata?
- GA implicit parallelism
- Schema Theorem with only selection
- Destructive and Constructive Effects of Crossover and Mutation on Schemata
- Probabilities of Schema Surviving Crossover and Surviving Mutation
-
Complete Schema Theorem (selection, crossover and mutation)
- Materials:
- Deadlines:
- GA Homework Assignment Due: October 17th, 2006
Lecture 5, October 5, 2006
- Evaluation (mini-test) covering the four previous lectures including all material handed out in class