Are Labels Necessary for Neural Architecture Search?

TL;DR

This paper introduces Unsupervised Neural Architecture Search (UnNAS), demonstrating that high-quality neural network architectures can be discovered using only images without labels, challenging the traditional reliance on labeled data.

Contribution

The paper defines UnNAS and provides empirical evidence that unsupervised objectives can produce architectures comparable to supervised methods, reducing dependence on labeled datasets.

Findings

Architecture rankings are highly correlated with and without labels.

Unsupervised NAS can find competitive architectures.

Labels may not be necessary for effective neural architecture search.

Abstract

Existing neural network architectures in computer vision -- whether designed by humans or by machines -- were typically found using both images and their associated labels. In this paper, we ask the question: can we find high-quality neural architectures using only images, but no human-annotated labels? To answer this question, we first define a new setup called Unsupervised Neural Architecture Search (UnNAS). We then conduct two sets of experiments. In sample-based experiments, we train a large number (500) of diverse architectures with either supervised or unsupervised objectives, and find that the architecture rankings produced with and without labels are highly correlated. In search-based experiments, we run a well-established NAS algorithm (DARTS) using various unsupervised objectives, and report that the architectures searched without labels can be competitive to their counterparts searched with labels. Together, these results reveal the potentially surprising finding that labels are not necessary, and the image statistics alone may be sufficient to identify good neural architectures.