Baudiš, P. - Pošík, P.: Global Line Search Algorithm Hybridized with Quadratic Interpolation and Its Extension to Separable Functions. In GECCO '15 Proceedings of the 2015 Annual Conference on Genetic and Evolutionary Computation. New York: ACM Press, 2015, vol. 1, p. 257-264. ISBN 978-1-4503-3472-3.
We propose a novel hybrid algorithm“Brent-STEP” for uni-
variate global function minimization, based on the global
line search method STEP and accelerated by Brent’s method,
a local optimizer that combines quadratic interpolation and
golden section steps. We analyze the performance of the hy-
brid algorithm on various one-dimensional functions and ex-
perimentally demonstrate a significant improvement relative
to its constituent algorithms in most cases. We then gener-
alize the algorithm to multivariate functions, adopting the
recently proposed [8] scheme to interleave evaluations across
dimensions to achieve smoother and more efficient conver-
gence. We experimentally demonstrate the highly competi-
tive performance of the proposed multivariate algorithm on
separable functions of the BBOB benchmark. The combina-
tion of good performance and smooth convergence on sepa-
rable functions makes the algorithm an interesting candidate
for inclusion in algorithmic portfolios or hybrid algorithms
that aim to provide good performance on a wide range of
problems.
en
dc.language.iso
en
en
dc.publisher
ACM
cze
dc.subject
Black-box optimization
en
dc.subject
Line search
en
dc.subject
Separable functions
en
dc.subject
Hybrid algorithm
en
dc.title
Global Line Search Algorithm Hybridized with Quadratic Interpolation and Its Extension to Separable Functions