Multiobjective Evolution for Convolutional Neural Network Architecture Search

0534830 - ÚI 2021 RIV CH eng C - Konferenční příspěvek (zahraniční konf.)
Vidnerová, Petra - Procházka, Štěpán - Neruda, Roman
Multiobjective Evolution for Convolutional Neural Network Architecture Search.
Artificial Intelligence and Soft Computing. ICAISC 2020 Proceedings, Part I. Cham: Springer, 2020 - (Rutkowski, L.; Scherer, R.; Korytkowski, M.; Pedrycz, W.; Tadeusiewicz, R.; Zurada, J.), s. 261-270. Lecture Notes on Computer Science, 12415. ISBN 978-3-030-61400-3. ISSN 0302-9743.
[ICAISC 2020: International Conference on Artificial Intelligence and Soft Computing /19./. Zakopane (PL), 12.10.2020-14.10.2020]
Grant CEP: GA ČR(CZ) GA18-23827S
Institucionální podpora: RVO:67985807
Klíčová slova: Multi-objective evolution * Neural architecture search * Convolutional neural networks
Obor OECD: Computer sciences, information science, bioinformathics (hardware development to be 2.2, social aspect to be 5.8)

The choice of an architecture is crucial for the performance of the neural network, and thus automatic methods for architecture search have been proposed to provide a data-dependent solution to this problem. In this paper, we deal with an automatic neural architecture search for convolutional neural networks. We propose a novel approach for architecture selection based on multi-objective evolutionary optimisation. Our algorithm optimises not only the performance of the network, but it controls also the size of the network, in terms of the number of network parameters. The proposed algorithm is evaluated on experiments, including MNIST and fashionMNIST classification problems. Our approach outperforms both the considered baseline architectures and the standard genetic algorithm.
Trvalý link: http://hdl.handle.net/11104/0312994