Probabilistic Neural Architecture Search

TL;DR

This paper introduces PARSEC, a probabilistic neural architecture search method that significantly reduces memory usage and computational cost, enabling efficient search over complex architectures and larger datasets.

Contribution

PARSEC presents a memory-efficient probabilistic framework for neural architecture search that transfers learned distributions across different dataset sizes.

Findings

Outperforms methods with double the computational cost on CIFAR-10.

Matches performance of methods with 1000x higher computational cost on ImageNet.

Requires only as much memory as training a single architecture.

Abstract

In neural architecture search (NAS), the space of neural network architectures is automatically explored to maximize predictive accuracy for a given task. Despite the success of recent approaches, most existing methods cannot be directly applied to large scale problems because of their prohibitive computational complexity or high memory usage. In this work, we propose a Probabilistic approach to neural ARchitecture SEarCh (PARSEC) that drastically reduces memory requirements while maintaining state-of-the-art computational complexity, making it possible to directly search over more complex architectures and larger datasets. Our approach only requires as much memory as is needed to train a single architecture from our search space. This is due to a memory-efficient sampling procedure wherein we learn a probability distribution over high-performing neural network architectures. Importantly, this framework enables us to transfer the distribution of architectures learnt on smaller problems to larger ones, further reducing the computational cost. We showcase the advantages of our approach in applications to CIFAR-10 and ImageNet, where our approach outperforms methods with double its computational cost and matches the performance of methods with costs that are three orders of magnitude larger.