We propose a neuroevolutionary speciation mechanism that is applied on NeuroEvolution of Augmenting Topologies (NEAT) that solely evolves neural networks{ extquoteright} topology and weights and its extension HA-NEAT that also evolves activation functions. The new speciation mechanism is defined based on the behavior of the individuals rather than their topological similarity. Focusing on classification tasks we build artificial datasets of high complexity. Performance is described by (i) median classification accuracy, (ii) computational efficiency (number of generations) and (iii) network complexity (number of nodes and connections). The performance metrics are compared using Kruskal-Wallis hypothesis tests with Bonferroni correction. It is found that the proposed behavioral speciation mechanism outperforms the original speciation solving problems that were not solvable before or improving the accuracy and reducing the network complexity
Papavasileiou, E, Cornelis, JPH & Jansen, B 2020, Behavior-based Speciation in Classification with NeuroEvolution. in 2020 IEEE World Congress on Computational Intelligence: 2020 IEEE Congress on Evolutionary Computation (CEC)., 9185720, 2020 IEEE Congress on Evolutionary Computation, CEC 2020 - Conference Proceedings, pp. 1-7, 2020 IEEE World Congress on Computational Intelligence , Glasgow, United Kingdom, 19/07/20. https://doi.org/10.1109/CEC48606.2020.9185720
Papavasileiou, E., Cornelis, J. P. H., & Jansen, B. (2020). Behavior-based Speciation in Classification with NeuroEvolution. In 2020 IEEE World Congress on Computational Intelligence: 2020 IEEE Congress on Evolutionary Computation (CEC) (pp. 1-7). Article 9185720 (2020 IEEE Congress on Evolutionary Computation, CEC 2020 - Conference Proceedings). https://doi.org/10.1109/CEC48606.2020.9185720
@inproceedings{b72eab6dc796480e9aa0b9bc309f57e5,
title = "Behavior-based Speciation in Classification with NeuroEvolution",
abstract = "We propose a neuroevolutionary speciation mechanism that is applied on NeuroEvolution of Augmenting Topologies (NEAT) that solely evolves neural networks{ extquoteright} topology and weights and its extension HA-NEAT that also evolves activation functions. The new speciation mechanism is defined based on the behavior of the individuals rather than their topological similarity. Focusing on classification tasks we build artificial datasets of high complexity. Performance is described by (i) median classification accuracy, (ii) computational efficiency (number of generations) and (iii) network complexity (number of nodes and connections). The performance metrics are compared using Kruskal-Wallis hypothesis tests with Bonferroni correction. It is found that the proposed behavioral speciation mechanism outperforms the original speciation solving problems that were not solvable before or improving the accuracy and reducing the network complexity",
author = "Evgenia Papavasileiou and Cornelis, {Jan Paul Herman} and Bart Jansen",
year = "2020",
month = jul,
doi = "10.1109/CEC48606.2020.9185720",
language = "English",
isbn = "978-1-7281-6930-9",
series = "2020 IEEE Congress on Evolutionary Computation, CEC 2020 - Conference Proceedings",
pages = "1--7",
booktitle = "2020 IEEE World Congress on Computational Intelligence",
note = "2020 IEEE World Congress on Computational Intelligence : 2020 IEEE Congress on Evolutionary Computation (CEC), IEEE WCCI 2020 ; Conference date: 19-07-2020 Through 24-07-2020",
}