Analyzing synchronous and asynchronous parallel distributed genetic algorithms
Future Generation Computer Systems - Special issue on bio-impaired solutions to parallel processing problems
Completely Derandomized Self-Adaptation in Evolution Strategies
Evolutionary Computation
Fitness Expectation Maximization
Proceedings of the 10th international conference on Parallel Problem Solving from Nature: PPSN X
Stochastic search using the natural gradient
ICML '09 Proceedings of the 26th Annual International Conference on Machine Learning
Exponential natural evolution strategies
Proceedings of the 12th annual conference on Genetic and evolutionary computation
High dimensions and heavy tails for natural evolution strategies
Proceedings of the 13th annual conference on Genetic and evolutionary computation
Speedups between ×70 and ×120 for a generic local search (memetic) algorithm on a single GPGPU chip
EvoApplicatons'10 Proceedings of the 2010 international conference on Applications of Evolutionary Computation - Volume Part I
| Hi-index | 0.00 |
We propose a generic method for turning a modern, non-elitist evolution strategy with fully adaptive covariance matrix into an asynchronous algorithm. This algorithm can process the result of an evaluation of the fitness function anytime and update its search strategy, without the need to synchronize with the rest of the population. The asynchronous update builds on the recent developments of natural evolution strategies and information geometric optimization. Our algorithm improves on the usual generational scheme in two respects. Remarkably, the possibility to process fitness values immediately results in a speed-up of the sequential algorithm. Furthermore, our algorithm is much better suited for parallel processing. It allows to use more processors than offspring individuals in a meaningful way.