Abstract
AbstractNSGA-Net is a popular method for neural architecture search (NAS). It conducts the improved non-dominated sorting genetic algorithm (NSGA-II) during its search procedure. In this paper, a NAS method using the multi-objective evolutionary algorithm based on decomposition (MOEA/D-Net) is proposed to heighten the running efficiency of NSGA-Net. MOEA/D-Net aims to minimize the number of floating-point operations (FLOPs) and error rate of neural architectures through the multi-objective evolutionary algorithm based on decomposition (MOEA/D) during the search process. It selects parents within the neighborhoods of a subproblem and conducts multi-point crossover and mutation to generate offspring individuals at every generation. Experiment results on the CIFAR-10 image classification dataset indicate that MOEA/D-Net obtains architecture networks with less FLOPs and MOEA/D-Net outperforms NSGA-Net in terms of running efficiency.KeywordsNeural architecture searchImage classificationMulti-objective optimizationEvolutionary algorithmsDecomposition
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
