K-shot NAS: Learnable Weight-Sharing for NAS with K-shot Supernets

TL;DR

This paper introduces K-shot supernets and simplex-net to improve neural architecture search by allowing adaptive, path-specific weight sharing, leading to more accurate evaluations and better overall performance.

Contribution

It proposes a novel K-shot supernet framework with a simplex-net for adaptive weight sharing, enhancing NAS evaluation accuracy and performance.

Findings

K-shot supernets outperform single supernets in NAS evaluation accuracy.

The approach achieves significant performance improvements on benchmark datasets.

Adaptive weight sharing enables more reliable architecture evaluation.

Abstract

In one-shot weight sharing for NAS, the weights of each operation (at each layer) are supposed to be identical for all architectures (paths) in the supernet. However, this rules out the possibility of adjusting operation weights to cater for different paths, which limits the reliability of the evaluation results. In this paper, instead of counting on a single supernet, we introduce $K$-shot supernets and take their weights for each operation as a dictionary. The operation weight for each path is represented as a convex combination of items in a dictionary with a simplex code. This enables a matrix approximation of the stand-alone weight matrix with a higher rank ($K>1$). A \textit{simplex-net} is introduced to produce architecture-customized code for each path. As a result, all paths can adaptively learn how to share weights in the $K$-shot supernets and acquire corresponding weights for better evaluation. $K$-shot supernets and simplex-net can be iteratively trained, and we further extend the search to the channel dimension. Extensive experiments on benchmark datasets validate that K-shot NAS significantly improves the evaluation accuracy of paths and thus brings in impressive performance improvements.