Título: Evolutionary Convolutional Neural Network: a case study
Autores: Amanda Lucas Pereira, Manoela Kohler and Marco Aurelio C. Pacheco
Resumo:
Most of the state-of-the-art CNNs architecture are manually crafted by experts, usually with background knowledge from extent working experience in this research field. Therefore, this manner of designing CNNs is highly limited and many approaches have been developed in trying to make this procedure more automatic. In this paper, it is presented a case study in tackling the architecture search problem by using a Genetic Algorithm (GA) in the optimization of an existing CNN Architecture. The proposed methodology uses VGG-16 convolutional blocks as its building blocks. Each individual from the GA corresponds to a possible model built on these building blocks with varying filter sizes, but keeping fixed the original network architecture connections. The selection of the fittest individuals are done according to their weighted F1-Score when training from scratch in the available data. To evaluate the best individual found from the proposed methodology, the performance is compared to a VGG-16 model trained from scratch on the same data.
Palavras-chave:
genetic algorithm, neural network architecture search, deep learning, evolutionary algorithms.
Páginas: 6
Código DOI: 10.21528/CBIC2021-129
Artigo em pdf: CBIC_2021_paper_129.pdf
Arquivo BibTeX: CBIC_2021_129.bib