Hybrid self-attention NEAT: a novel evolutionary self-attention approach to improve the NEAT algorithm in high dimensional inputs

This article presents a "Hybrid Self-Attention NEAT" method to improve the original NeuroEvolution of Augmenting Topologies (NEAT) algorithm in high-dimensional inputs. Although NEAT has shown significant result different challenging tasks, as input representations are high dimensional, it cannot create well-tuned network. Our study addresses this limitation by using self-attention an indirect encoding select most important parts input. In addition, we its overall performance with help hybrid evolve final network weights. The main conclusion is that Hybrid Self- Attention can eliminate restriction NEAT. results indicate comparison evolutionary algorithms, our model get comparable scores Atari games raw pixels much lower number parameters.