Research breakthrough possible @S-Logix pro@slogix.in

Office Address

Social List

Comparing Large-Scale Global Optimization Competition winners in a real-world problem - 2019

Comparing Large-Scale Global OptimizationCompetition winners in a real-world problem

Research paper on Comparing Large-Scale Global Optimization Competition winners in a real-world problem

Research Area:  Metaheuristic Computing

Abstract:

The optimization of thousands of variables, Large-Scale Global Optimization, is a research topic that is obtaining more and more attention by its applications in engineering and medical problems. In order to design evolutionary algorithms for these problems, several specific competitions have been organized, using benchmarks such as the ones proposed in CEC-2010 and CEC-2013, trying to simulate realistic features of real-world problems. Several algorithms have been proposed, some of them being very competitive on these benchmarks, especially during the last years. However, all of them were tested only on those artificial benchmarks, so there are no guarantees that they would obtain good performance in more realistic problems. In this paper, we select the best algorithms in these competitions to optimize a real-world problem, an electroencephalography (EEG) optimization problem. The new benchmark contains noisy problems and an increasing number of variables (up to 5000) compared to synthetic benchmarks (limited to 1000 variables). Results show that, although the fitness obtained by the majority of the algorithms is the same, the processing time strongly depends on the algorithm under consideration. The optimization time for a fixed number of fitness evaluations varies, in the most complex problems, from 3 hours to around 18 minutes, being MOS-2013 the fastest algorithm. However, if we focus our attention on the time needed to reach the best-known solution, SHADEILS becomes the fastest algorithm (with a maximum of three minutes). In our opinion, this should encourage researchers to continue working in more scalable and efficient algorithms for large-scale global optimization.

Keywords:  
Large-scale global optimization
LSGO
Benchmarking
BComp
electroencephalography (EEG)
Meta-Heuristic

Author(s) Name:  Daniel Molina; Arthur R. Nesterenko; Antonio LaTorre

Journal name:  

Conferrence name:  2019 IEEE Congress on Evolutionary Computation (CEC)

Publisher name:  IEEE

DOI:  10.1109/CEC.2019.8789943

Volume Information: