Multi-Objective Reinforced Evolution in Mobile Neural Architecture Search

TL;DR

This paper introduces MoreMNAS, a multi-objective neural architecture search method combining evolutionary algorithms and reinforcement learning to optimize mobile neural networks across multiple criteria.

Contribution

It presents a novel multi-objective genetic algorithm that integrates reinforcement learning to enhance search efficiency and model quality for mobile neural networks.

Findings

Achieves competitive super-resolution models with fewer FLOPS.

Balances exploration and exploitation effectively during search.

Outperforms some state-of-the-art methods in super-resolution domain.

Abstract

Fabricating neural models for a wide range of mobile devices demands for a specific design of networks due to highly constrained resources. Both evolution algorithms (EA) and reinforced learning methods (RL) have been dedicated to solve neural architecture search problems. However, these combinations usually concentrate on a single objective such as the error rate of image classification. They also fail to harness the very benefits from both sides. In this paper, we present a new multi-objective oriented algorithm called MoreMNAS (Multi-Objective Reinforced Evolution in Mobile Neural Architecture Search) by leveraging good virtues from both EA and RL. In particular, we incorporate a variant of multi-objective genetic algorithm NSGA-II, in which the search space is composed of various cells so that crossovers and mutations can be performed at the cell level. Moreover, reinforced control is mixed with a natural mutating process to regulate arbitrary mutation, maintaining a delicate balance between exploration and exploitation. Therefore, not only does our method prevent the searched models from degrading during the evolution process, but it also makes better use of learned knowledge. Our experiments conducted in Super-resolution domain (SR) deliver rivalling models compared to some state-of-the-art methods with fewer FLOPS.