Debiased Self-Training for Semi-Supervised Learning

TL;DR

This paper introduces Debiased Self-Training (DST), a novel method that reduces bias and improves stability in semi-supervised learning by decoupling pseudo label generation and adversarially optimizing representations.

Contribution

The paper proposes DST, which decouples pseudo label generation from training and adversarially optimizes representations to mitigate bias and improve semi-supervised learning performance.

Findings

DST achieves 6.3% average improvement over state-of-the-art methods.

DST improves stability and class balance in training.

DST enhances performance when training from scratch or fine-tuning.

Abstract

Deep neural networks achieve remarkable performances on a wide range of tasks with the aid of large-scale labeled datasets. Yet these datasets are time-consuming and labor-exhaustive to obtain on realistic tasks. To mitigate the requirement for labeled data, self-training is widely used in semi-supervised learning by iteratively assigning pseudo labels to unlabeled samples. Despite its popularity, self-training is well-believed to be unreliable and often leads to training instability. Our experimental studies further reveal that the bias in semi-supervised learning arises from both the problem itself and the inappropriate training with potentially incorrect pseudo labels, which accumulates the error in the iterative self-training process. To reduce the above bias, we propose Debiased Self-Training (DST). First, the generation and utilization of pseudo labels are decoupled by two parameter-independent classifier heads to avoid direct error accumulation. Second, we estimate the worst case of self-training bias, where the pseudo labeling function is accurate on labeled samples, yet makes as many mistakes as possible on unlabeled samples. We then adversarially optimize the representations to improve the quality of pseudo labels by avoiding the worst case. Extensive experiments justify that DST achieves an average improvement of 6.3% against state-of-the-art methods on standard semi-supervised learning benchmark datasets and 18.9%$ against FixMatch on 13 diverse tasks. Furthermore, DST can be seamlessly adapted to other self-training methods and help stabilize their training and balance performance across classes in both cases of training from scratch and finetuning from pre-trained models.