Learning from Experience: Case Injected Genetic Algorithm Design of Combinational Logic Circuits Sushil J. Louis Genetic Algorithm Systems Lab(gaslab) University of Nevada Reno
Outline Background What is the technique? GAs + CBR How do we evaluate the technique? Example problem from Combinational Logic Design Is the technique useful? Results Conclusions
Background Genetic Algorithm augmentation Deployed systems are expected to confront and solve many problems over their lifetime How can we increase genetic algorithm performance with experience? Provide GA with a memory Seed the GA population
Case-Based Reasoning When confronted by a new problem, adapt similar (already solved) problem’s solution to solve new problem Many problems in design are suited to a case- based representation CBR = Associative Memory + Adaptation Indexing (similarity) and adaptation are domain dependent
Case Injected Genetic AlgoRithm Combine genetic “adaptive” search with case-based memory Case-base provides memory Genetic algorithm provides adaptation Questions: What is a case? How do we do Indexing?
What is a Case? CIGAR Member of the GA’s population (Chromosome) Fitness Generation that this chromosome was created Other
Indexing Problem similarity We must have a similarity metric over problems Solution similarity We use hamming distance for binary encodings, sequence similarity for permutation encodings.
Problem Similarity
Solution Similarity