CS 478 - Evolutionary Algorithms 1 - Neural Networks and Machine ...

More documents

Recommendations

Info

Start with initialized population P(t) - random, domainknowledge, etc. Typically have fixed population size (type of beam search), large enough to maintain diversity Selection – Parent_Selection P(t) - Promising Parents more likely to be chosen based on fitness to create new children using genetic operators – Survive P(t) - Pruning of less promising candidates, Evaluate P(t) - Calculate fitness of population members. Could be simple metrics to complex simulations. Survival of the fittest - Find and keep best while also maintaining diversity CS 478 - Evolutionary Algorithms 4
Procedure EA t = 0; Initialize Population P(t); Evaluate P(t); Until Done{ /*Sufficiently “good” individuals discovered or many iterations passed with no improvement, etc.*/ t = t + 1; Parent_Selection P(t); Recombine P(t); Mutate P(t); Evaluate P(t); Survive P(t);} CS 312 – Approximation 5
Page 1 and 2: CS 478 - Evolutionary Algorithms 1
Page 3: Populate our search space with init
Page 9 and 10: Individuals are represented so that
Page 13 and 14: In general want the fittest parents
Page 15 and 16: Population size - Larger gives more
Page 17 and 18: There exist mathematical proofs tha
Page 19 and 20: Representation that best fits probl
Page 21 and 22: Evolves more complex structures - p
Page 23 and 24: Use a Genetic Algorithm to learn th
Page 25 and 26: Much current work and extensions Nu
Page 27 and 28: Each classifier has an associated s
Page 29: s(C,t+1) = s(C,t) -B(C,t) - Loss of

CS 478 - Evolutionary Algorithms 1 - Neural Networks and Machine ...

Create successful ePaper yourself

Delete template?

Save as template?