Attention Based Evolutionary Approach for Image Classification

EasyChair Preprint 9574

Abstract

Lately, evolutionary algorithms have gained traction due to their ability to produce state-of-the-art deep learning architectures for a given data set, even though they require considerable amount of compute resources, they are a heavily researched domain because of the complexities involved in designing deep learning architectures. Currently, none of the evolutionary approaches available have incorporated the attention mechanism, which is a proven technique to improve the performance of image classification and language models. This paper posits a neuroevolutionary technique coupled with the use of Convolution Block Attention Module for image classification. As technology progresses, it’s inevitable that there will be massive advancements leading to cheaper and more available computing making evolutionary approaches a promising avenue to develop task specific deep learning models. The proposed approach evolves a topology that achieves a high fitness of 87.44% on the CIFAR-10 benchmark, using fewer parameters as compared to previous approaches. This results in a superior fitness score compared to most past approaches, despite being evolved for just few generations.

Keyphrases: CoDeepNEAT Convolutional Block AttentionModule(CBAM) CIFAR-10, Genetic Algorithms, Neuro-evolution, attention, topology evolution

@booklet{EasyChair:9574,
  author    = {Ajay Prem and Anirudh Joshi and Haritha Madana and Jaywanth Gowda and Arti Arya},
  title     = {Attention Based Evolutionary Approach for Image Classification},
  howpublished = {EasyChair Preprint 9574},
  year      = {EasyChair, 2023}}