USB: A Unified Semi-supervised Learning Benchmark for Classification

TL;DR

This paper introduces a comprehensive benchmark, USB, for evaluating semi-supervised learning across multiple domains, providing a unified, efficient, and open-source platform for fair comparison and pre-trained models.

Contribution

The paper presents a new unified SSL benchmark covering CV, NLP, and Audio, with a modular codebase and pre-trained models, enabling efficient cross-domain evaluation.

Findings

USB reduces evaluation cost significantly.

It enables evaluation of SSL methods across multiple domains.

Provides pre-trained models for CV tasks.

Abstract

Semi-supervised learning (SSL) improves model generalization by leveraging massive unlabeled data to augment limited labeled samples. However, currently, popular SSL evaluation protocols are often constrained to computer vision (CV) tasks. In addition, previous work typically trains deep neural networks from scratch, which is time-consuming and environmentally unfriendly. To address the above issues, we construct a Unified SSL Benchmark (USB) for classification by selecting 15 diverse, challenging, and comprehensive tasks from CV, natural language processing (NLP), and audio processing (Audio), on which we systematically evaluate the dominant SSL methods, and also open-source a modular and extensible codebase for fair evaluation of these SSL methods. We further provide the pre-trained versions of the state-of-the-art neural models for CV tasks to make the cost affordable for further tuning. USB enables the evaluation of a single SSL algorithm on more tasks from multiple domains but with less cost. Specifically, on a single NVIDIA V100, only 39 GPU days are required to evaluate FixMatch on 15 tasks in USB while 335 GPU days (279 GPU days on 4 CV datasets except for ImageNet) are needed on 5 CV tasks with TorchSSL.