Autors: Tsokov, S. A., Lazarova, M. K., AleksievaPetrova, A. P.
Title: Optimizing Neural Network Architectures Using an Evolutionary Algorithm with a Modified Crossover Operator
Keywords: autonomous agent, convolutional neural networks, deep learning, genetic algorithms, human activity recognition, neural network architecture design, neuroevolution

Abstract: The performance of neural networks can depend a lot on their architecture, the optimization of which is usually a difficult task. This article presents a genetic algorithm for optimizing neural network architectures, which gradually complexifies the networks in the course of the search. The method uses a modified operator to cross solutions with varying lengths. The method was used to optimize two different types of neural networks, solving two different problems, and the results show that the obtained architectures performed well.

References

    Issue

    Journal Computer and Communication Engineering, vol. 15, issue 1, 2021, Bulgaria, ISSN 1314-2291

    Вид: статия в списание, публикация в реферирано издание