Neural Architecture Performance Prediction Using Graph Neural Networks

TL;DR

This paper introduces a Graph Neural Network-based surrogate model for predicting neural architecture performance, enabling zero-shot predictions on unseen architectures in NAS, reducing computational costs.

Contribution

The paper presents a novel GNN-based surrogate model for neural architecture performance prediction, addressing the challenge of predicting for unknown architectures in NAS.

Findings

Effective zero-shot prediction on NAS-Bench-101

Outperforms existing surrogate models

Reduces computational costs in NAS

Abstract

In computer vision research, the process of automating architecture engineering, Neural Architecture Search (NAS), has gained substantial interest. Due to the high computational costs, most recent approaches to NAS as well as the few available benchmarks only provide limited search spaces. In this paper we propose a surrogate model for neural architecture performance prediction built upon Graph Neural Networks (GNN). We demonstrate the effectiveness of this surrogate model on neural architecture performance prediction for structurally unknown architectures (i.e. zero shot prediction) by evaluating the GNN on several experiments on the NAS-Bench-101 dataset.