Semi-Supervised Learning via Weight-aware Distillation under Class Distribution Mismatch

TL;DR

This paper introduces Weight-Aware Distillation (WAD), a novel semi-supervised learning framework designed to effectively handle class distribution mismatch by selectively transferring knowledge and filtering unknown categories, backed by theoretical guarantees and empirical results.

Contribution

The paper proposes WAD, a new SSL method that uses adaptive weights and pseudo labels to mitigate class mismatch issues, with proven risk bounds and superior performance.

Findings

WAD outperforms five state-of-the-art SSL methods on benchmark datasets.

Theoretical proof of tight population risk bound under class mismatch.

Effective filtering of unknown categories improves SSL accuracy.

Abstract

Semi-Supervised Learning (SSL) under class distribution mismatch aims to tackle a challenging problem wherein unlabeled data contain lots of unknown categories unseen in the labeled ones. In such mismatch scenarios, traditional SSL suffers severe performance damage due to the harmful invasion of the instances with unknown categories into the target classifier. In this study, by strict mathematical reasoning, we reveal that the SSL error under class distribution mismatch is composed of pseudo-labeling error and invasion error, both of which jointly bound the SSL population risk. To alleviate the SSL error, we propose a robust SSL framework called Weight-Aware Distillation (WAD) that, by weights, selectively transfers knowledge beneficial to the target task from unsupervised contrastive representation to the target classifier. Specifically, WAD captures adaptive weights and high-quality pseudo labels to target instances by exploring point mutual information (PMI) in representation space to maximize the role of unlabeled data and filter unknown categories. Theoretically, we prove that WAD has a tight upper bound of population risk under class distribution mismatch. Experimentally, extensive results demonstrate that WAD outperforms five state-of-the-art SSL approaches and one standard baseline on two benchmark datasets, CIFAR10 and CIFAR100, and an artificial cross-dataset. The code is available at https://github.com/RUC-DWBI-ML/research/tree/main/WAD-master.