Permuted Orthogonal Block-Diagonal Transformation Matrices for Large Scale Optimization Benchmarking - Inria - Institut national de recherche en sciences et technologies du numérique Access content directly
Conference Papers Year : 2016

Permuted Orthogonal Block-Diagonal Transformation Matrices for Large Scale Optimization Benchmarking

Ouassim Ait Elhara
  • Function : Author
  • PersonId : 938167
Anne Auger
  • Function : Author
  • PersonId : 751513
  • IdHAL : anne-auger
Nikolaus Hansen

Abstract

We propose a general methodology to construct large-scale testbeds for the benchmarking of continuous optimization algorithms. Our approach applies an orthogonal transformation on raw functions that involve only a linear number of operations in order to obtain large scale optimization benchmark problems. The orthogonal transformation is sampled from a parametrized family of transformations that are the product of a permutation matrix times a block-diagonal matrix times a permutation matrix. We investigate the impact of the different parameters of the transformation on its shape and on the difficulty of the problems for separable CMA-ES. We illustrate the use of the above defined transformation in the BBOB-2009 testbed as replacement for the expensive orthogonal (rotation) matrices. We also show the practicability of the approach by studying the computational cost and its applicability in a large scale setting.
Fichier principal
Vignette du fichier
permutedBlockDiagonalOrthogonalTransfromationMatrix.pdf (876.29 Ko) Télécharger le fichier
Origin Publisher files allowed on an open archive
Loading...

Dates and versions

hal-01308566 , version 1 (28-04-2016)
hal-01308566 , version 2 (02-05-2016)
hal-01308566 , version 3 (11-05-2016)

Identifiers

Cite

Ouassim Ait Elhara, Anne Auger, Nikolaus Hansen. Permuted Orthogonal Block-Diagonal Transformation Matrices for Large Scale Optimization Benchmarking. GECCO 2016, Jul 2016, Denver, United States. pp.189-196, ⟨10.1145/2908812.2908937⟩. ⟨hal-01308566v3⟩
405 View
723 Download

Altmetric

Share

Gmail Mastodon Facebook X LinkedIn More