GPU-Based Automatic Configuration of Differential Evolution: A Case Study

The performance of an evolutionary algorithm strongly depends on the choice of the parameters which regulate its behavior. In this paper, two evolutionary algorithms (Particle Swarm Optimization and Differential Evolution) are used to find the optimal configuration of parameters for Differential Evolution. We tested our approach on four benchmark functions, and the comparison with an exhaustive search demonstrated its effectiveness. Then, the same method was used to tune the parameters of Differential Evolution in solving a real-world problem: the automatic localization of the hippocampus in histological brain images. The results obtained consistently outperformed the ones achieved using manually-tuned parameters. Thanks to a GPU-based implementation, our tuner is up to 8 times faster than the corresponding sequential version.

[1]  Mario Giacobini,et al.  Automatic hippocampus localization in histological images using Differential Evolution-based deformable models , 2013, Pattern Recognit. Lett..

[2]  Renato A. Krohling,et al.  Swarm's flight: Accelerating the particles using C-CUDA , 2009, 2009 IEEE Congress on Evolutionary Computation.

[3]  A. E. Eiben,et al.  Comparing parameter tuning methods for evolutionary algorithms , 2009, 2009 IEEE Congress on Evolutionary Computation.

[4]  Jacek M. Zurada,et al.  Swarm and Evolutionary Computation , 2012, Lecture Notes in Computer Science.

[5]  James Kennedy,et al.  Particle swarm optimization , 1995, Proceedings of ICNN'95 - International Conference on Neural Networks.

[6]  John J. Grefenstette,et al.  Optimization of Control Parameters for Genetic Algorithms , 1986, IEEE Transactions on Systems, Man, and Cybernetics.

[7]  A. E. Eiben,et al.  Parameter tuning for configuring and analyzing evolutionary algorithms , 2011, Swarm Evol. Comput..

[8]  Renato A. Krohling,et al.  Differential evolution algorithm on the GPU with C-CUDA , 2010, IEEE Congress on Evolutionary Computation.

[9]  R. Storn,et al.  Differential Evolution - A simple and efficient adaptive scheme for global optimization over continuous spaces , 2004 .

[10]  A. E. Eiben,et al.  Beating the ‘world champion’ evolutionary algorithm via REVAC tuning , 2010, IEEE Congress on Evolutionary Computation.

[11]  Robert E. Mercer,et al.  ADAPTIVE SEARCH USING A REPRODUCTIVE META‐PLAN , 1978 .

[12]  A. E. Eiben,et al.  Introduction to Evolutionary Computing , 2003, Natural Computing Series.

[13]  Stefano Cagnoni,et al.  libCudaOptimize: an open source library of GPU-based metaheuristics , 2012, GECCO '12.

[14]  Gisbert Schneider,et al.  Optimized Particle Swarm Optimization (OPSO) and its application to artificial neural network training , 2006, BMC Bioinformatics.

[15]  M. E. H. Pedersen,et al.  Tuning & simplifying heuristical optimization , 2010 .

[16]  P. N. Suganthan,et al.  Differential Evolution: A Survey of the State-of-the-Art , 2011, IEEE Transactions on Evolutionary Computation.

[17]  Stefano Cagnoni,et al.  Algorithm configuration using GPU-based metaheuristics , 2013, GECCO '13 Companion.

[18]  Marco Dorigo,et al.  Swarm intelligence: from natural to artificial systems , 1999 .

[19]  Fabio Daolio,et al.  Evaluation of parallel particle swarm optimization algorithms within the CUDA™ architecture , 2011, Inf. Sci..

[20]  Václav Snásel,et al.  Many-threaded implementation of differential evolution for the CUDA platform , 2011, GECCO '11.

[21]  Stefano Cagnoni,et al.  GPU-based asynchronous particle swarm optimization , 2011, GECCO '11.