Diversify and Conquer: Open-set Disagreement for Robust Semi-supervised Learning with Outliers

TL;DR

This paper introduces Diversify and Conquer (DAC), a novel open-set semi-supervised learning framework that uses multiple biased models to detect outliers and improve robustness when unlabeled data contains unknown classes.

Contribution

DAC constructs multiple biased models within a single training process to exploit prediction disagreements for robust outlier detection in semi-supervised learning.

Findings

DAC improves outlier detection accuracy.

The method enhances SSL robustness against outliers.

Experimental results demonstrate superior performance over existing methods.

Abstract

Conventional semi-supervised learning (SSL) ideally assumes that labeled and unlabeled data share an identical class distribution, however in practice, this assumption is easily violated, as unlabeled data often includes unknown class data, i.e., outliers. The outliers are treated as noise, considerably degrading the performance of SSL models. To address this drawback, we propose a novel framework, Diversify and Conquer (DAC), to enhance SSL robustness in the context of open-set semi-supervised learning. In particular, we note that existing open-set SSL methods rely on prediction discrepancies between inliers and outliers from a single model trained on labeled data. This approach can be easily failed when the labeled data is insufficient, leading to performance degradation that is worse than naive SSL that do not account for outliers. In contrast, our approach exploits prediction disagreements among multiple models that are differently biased towards the unlabeled distribution. By leveraging the discrepancies arising from training on unlabeled data, our method enables robust outlier detection even when the labeled data is underspecified. Our key contribution is constructing a collection of differently biased models through a single training process. By encouraging divergent heads to be differently biased towards outliers while making consistent predictions for inliers, we exploit the disagreement among these heads as a measure to identify unknown concepts. Our code is available at https://github.com/heejokong/DivCon.