Evolutionary Algorithm- they make use of selection mutation and recombination, to build better population of solutions.

Genetic Algorithm that processes fixed length strings

• Selection
• Cross-over
• Mutation

Tournaments to decide

A = 11111

B = 00000

A = 11111

B = 00000

A’ = 00011

B’ = 11100

Assume all the previous one’s are good, but what if the following happens?

A = 00111

B = 11100

Path1 = Left1 + Right1

Path2 = Left2 + Right2

Path3 = Left1 + Right2

Path4 = Left2 + Right1

Tanimoto text has about 6-7 pages on a solution for the traveling salesperson problem (TSP)

• Remove "arbitrarily" one city replaced with "another"
• exchange arbitrary cities in the path

Tanimoto

Pseuodocode p. 242

LISP code p. 247

Setup initial population for g generations

• create m mutations
• add to the population
• check constraints & mutations
• create crossovers
• add to the population
• check crossover against constraints
• selection one (crossover = innovation)
• selection & mutation = parallel, noise tolerant, hill climbing
• selection & mutation & crossover = guard against losing diversity as a hill climber
• work good with samples
• word good with little knowledge

Class text books have LISP and C++ examples on Expert systems