Analyses of Evolution Strategy Behaviour
Evolution strategies are nature inspired algorithms for black box
optimisation. Mathematical analyses of their behaviour yield scaling laws
that aid practitioners in the choice of strategy variants, the setting of
their parameters, and the design of better algorithms. Adaptive variants
can be compared with each other as well as with hypothetical optimal
behaviour. An ongoing challenge is to extend the boundaries of the analyses
to include both strategies and test function classes of increasing
complexity. Our recent work considers a range of step size adaptation
mechanisms as well as several convex-quadratic and ridge functions and
noisy, time varying, and constrained problems.
Publications
D. V. Arnold, H.-G. Beyer, and A. Melkozerov
On the behaviour
of weighted multi-recombination evolution strategies optimising noisy
cigar functions
Genetic and Evolutionary Computation Conference, Montreal, 2009.
D. V. Arnold and D. Brauer
On the
behaviour of the (1+1)-ES for a simple constrained problem
Parallel Problem Solving from Nature — PPSN X, Dortmund, 2008.
D. V. Arnold and A. MacLeod
Step length
adaptation on ridge functions
Evolutionary Computation, 16(2):151-184, 2008.
D. V. Arnold
On the use of
evolution strategies for optimising certain positive definite quadratic
forms
Genetic and Evolutionary Computation Conference, London, 2007.
D. V. Arnold and H.-G. Beyer
Optimum
tracking with evolution strategies
Evolutionary Computation, 14(3):291-308, 2006.
D. V. Arnold and H.-G. Beyer
A general noise
model and its effects on evolution strategy performance
IEEE Transactions on Evolutionary Computation, 10(4):380-391, 2006.
D. V. Arnold and H.-G. Beyer
Performance
analysis of evolutionary optimization with cumulative step length
adaptation
IEEE Transactions on Automatic Control, 49(4):617-622, 2004.
D. V. Arnold
Noisy Optimization with Evolution
Strategies
Kluwer Academic Publishers, 2002.
Support
This research is supported through grants from the Natural Sciences and
Engineering Research Council of Canada (NSERC) and the Canada Foundation
for Innovation (CFI).
|