Efficient Multi-objective Neural Architecture Search via Lamarckian Evolution

TL;DR

This paper introduces LEMONADE, an evolutionary algorithm for multi-objective neural architecture search that efficiently finds architectures balancing performance and resource use, using Lamarckian inheritance to reduce computational costs.

Contribution

The paper proposes LEMONADE, a novel multi-objective evolutionary algorithm with Lamarckian inheritance for efficient neural architecture search under resource constraints.

Findings

LEMONADE effectively approximates the Pareto-front of architectures.

The Lamarckian inheritance mechanism accelerates search by warmstarting child networks.

Models found by LEMONADE outperform some hand-crafted and automatic architectures.

Abstract

Neural Architecture Search aims at automatically finding neural architectures that are competitive with architectures designed by human experts. While recent approaches have achieved state-of-the-art predictive performance for image recognition, they are problematic under resource constraints for two reasons: (1)the neural architectures found are solely optimized for high predictive performance, without penalizing excessive resource consumption, (2) most architecture search methods require vast computational resources. We address the first shortcoming by proposing LEMONADE, an evolutionary algorithm for multi-objective architecture search that allows approximating the entire Pareto-front of architectures under multiple objectives, such as predictive performance and number of parameters, in a single run of the method. We address the second shortcoming by proposing a Lamarckian inheritance mechanism for LEMONADE which generates children networks that are warmstarted with the predictive performance of their trained parents. This is accomplished by using (approximate) network morphism operators for generating children. The combination of these two contributions allows finding models that are on par or even outperform both hand-crafted as well as automatically-designed networks.