MultiMatch: Multihead Consistency Regularization Matching for Semi-Supervised Text Classification

TL;DR

MultiMatch is a new semi-supervised text classification method that combines multiple consistency and agreement techniques, achieving state-of-the-art results and robustness on various NLP benchmarks.

Contribution

It introduces a pseudo-label weighting module that unifies and enhances existing SSL techniques, improving robustness and performance in semi-supervised NLP tasks.

Findings

Achieves state-of-the-art results on 8 out of 10 NLP benchmark setups.

Ranks first among 21 methods according to the Friedman test.

Outperforms competitors by 3.26% in highly imbalanced settings.

Abstract

We introduce MultiMatch, a novel semi-supervised learning (SSL) algorithm combining the paradigms of co-training and consistency regularization with pseudo-labeling. At its core, MultiMatch features a pseudo-label weighting module designed for selecting and filtering pseudo-labels based on head agreement and model confidence, and weighting them according to the perceived classification difficulty. This novel module enhances and unifies three existing techniques -- heads agreement from Multihead Co-training, self-adaptive thresholds from FreeMatch, and Average Pseudo-Margins from MarginMatch -- resulting in a holistic approach that improves robustness and performance in SSL settings. Experimental results on benchmark datasets highlight the superior performance of MultiMatch, i.e., MultiMatch achieves state-of-the-art results on 8 out of 10 setups from 5 natural language processing datasets and ranks first according to the Friedman test among 21 methods. Furthermore, MultiMatch demonstrates exceptional robustness in highly imbalanced settings, outperforming the second-best approach by 3.26%, a critical advantage for real-world text classification tasks. Our code is available on GitHub.