A Channel-ensemble Approach: Unbiased and Low-variance Pseudo-labels is Critical for Semi-supervised Classification
This work addresses a critical bottleneck in semi-supervised classification for computer vision applications, offering a lightweight and extensible solution to enhance existing frameworks like FixMatch and FreeMatch.
The paper tackles the problem of biased and high-variance pseudo-labels in semi-supervised learning, especially with limited labeled data, by proposing a channel-based ensemble method that consolidates multiple inferior pseudo-labels into unbiased and low-variance ones, resulting in significant performance improvements on CIFAR10/100 datasets.
Semi-supervised learning (SSL) is a practical challenge in computer vision. Pseudo-label (PL) methods, e.g., FixMatch and FreeMatch, obtain the State Of The Art (SOTA) performances in SSL. These approaches employ a threshold-to-pseudo-label (T2L) process to generate PLs by truncating the confidence scores of unlabeled data predicted by the self-training method. However, self-trained models typically yield biased and high-variance predictions, especially in the scenarios when a little labeled data are supplied. To address this issue, we propose a lightweight channel-based ensemble method to effectively consolidate multiple inferior PLs into the theoretically guaranteed unbiased and low-variance one. Importantly, our approach can be readily extended to any SSL framework, such as FixMatch or FreeMatch. Experimental results demonstrate that our method significantly outperforms state-of-the-art techniques on CIFAR10/100 in terms of effectiveness and efficiency.