PercentMatch: Percentile-based Dynamic Thresholding for Multi-Label Semi-Supervised Classification

TL;DR

PercentMatch introduces a percentile-based dynamic thresholding method for multi-label semi-supervised classification, effectively leveraging unlabeled data and reducing noise, leading to improved performance on benchmark datasets.

Contribution

The paper proposes PercentMatch, a novel percentile-based threshold adjustment scheme for multi-label SSL, addressing challenges unique to multi-label tasks.

Findings

Achieves strong performance on Pascal VOC2007 and MS-COCO datasets.

Effectively reduces noise from early-stage unlabeled predictions.

Demonstrates advantages over recent SSL methods in multi-label classification.

Abstract

While much of recent study in semi-supervised learning (SSL) has achieved strong performance on single-label classification problems, an equally important yet underexplored problem is how to leverage the advantage of unlabeled data in multi-label classification tasks. To extend the success of SSL to multi-label classification, we first analyze with illustrative examples to get some intuition about the extra challenges exist in multi-label classification. Based on the analysis, we then propose PercentMatch, a percentile-based threshold adjusting scheme, to dynamically alter the score thresholds of positive and negative pseudo-labels for each class during the training, as well as dynamic unlabeled loss weights that further reduces noise from early-stage unlabeled predictions. Without loss of simplicity, we achieve strong performance on Pascal VOC2007 and MS-COCO datasets when compared to recent SSL methods.