Title :
A statistically aligned recombination operator for finite state machines
Author_Institution :
Univ. of Essex, Colchester
Abstract :
Learning finite state machines from samples of data has been extensively studied within machine learning and since the dawn of evolutionary computation. Conventional crossover or recombination operators used for finite state machines suffer from the competing conventions problem, caused by the combinatorial number of isomorphisms of each distinct machine. This paper introduces an efficient alignment operator to counteract this phenomenon. Results show that when in the neighbourhood of the target machine, the aligned crossover operator reaches the optimum in far few steps (on average) than either a naive crossover operator or a standard flip-style mutation operator.
Keywords :
combinatorial mathematics; evolutionary computation; finite state machines; learning (artificial intelligence); aligned crossover operator; combinatorial number; evolutionary computation; finite state machines; flip-style mutation operator; machine learning; statistically aligned recombination operator; Application software; Automata; Doped fiber amplifiers; Evolutionary computation; Genetic algorithms; Genetic mutations; Inference algorithms; Labeling; Machine learning; Recurrent neural networks; Finite state machines; alignment; crossover; recombination;
Conference_Titel :
Evolutionary Computation, 2007. CEC 2007. IEEE Congress on
Conference_Location :
Singapore
Print_ISBN :
978-1-4244-1339-3
Electronic_ISBN :
978-1-4244-1340-9
DOI :
10.1109/CEC.2007.4425068
