On the behavioral diversity of random programs

13 years 10 months ago

Download www.cs.bham.ac.uk

Generating a random sampling of program trees with speciﬁed function and terminal sets is the initial step of many program evolution systems. I present a theoretical and experimental analysis of the expected distribution of uniformly sampled programs, guided by algorithmic information theory. This analysis demonstrates that increasing the sample size is often an ineﬃcient means of increasing the overall diversity of program behaviors (outputs). A novel sampling scheme (semantic sampling) is proposed that exploits semantics to heuristically increase behavioral diversity. An important property of the scheme is that no calls of the problem-speciﬁc ﬁtness function are required. Its eﬀectiveness at increasing behavioral diversity is demonstrated empirically for Boolean formulae. Furthermore, it is found to lead to statistically signiﬁcant improvements in performance for genetic programming on parity and multiplexer problems. Categories and Subject Descriptors I.2.2 [Artiﬁcial...

Moshe Looks

Real-time Traffic