InterLUDE: Interactions between Labeled and Unlabeled Data to Enhance Semi-Supervised Learning

TL;DR

InterLUDE introduces a novel semi-supervised learning approach that leverages interactions between labeled and unlabeled data through embedding fusion and a new consistency-based loss, significantly improving classification accuracy.

Contribution

The paper proposes InterLUDE, a semi-supervised learning method that explicitly models interactions between labeled and unlabeled data, enhancing representation learning and prediction consistency.

Findings

Achieves 3.2% error on STL-10 with 40 labels, outperforming previous methods.

Demonstrates benefits on standard SSL benchmarks and medical SSL tasks.

Improves semi-supervised learning performance by modeling labeled-unlabeled interactions.

Abstract

Semi-supervised learning (SSL) seeks to enhance task performance by training on both labeled and unlabeled data. Mainstream SSL image classification methods mostly optimize a loss that additively combines a supervised classification objective with a regularization term derived solely from unlabeled data. This formulation neglects the potential for interaction between labeled and unlabeled images. In this paper, we introduce InterLUDE, a new approach to enhance SSL made of two parts that each benefit from labeled-unlabeled interaction. The first part, embedding fusion, interpolates between labeled and unlabeled embeddings to improve representation learning. The second part is a new loss, grounded in the principle of consistency regularization, that aims to minimize discrepancies in the model's predictions between labeled versus unlabeled inputs. Experiments on standard closed-set SSL benchmarks and a medical SSL task with an uncurated unlabeled set show clear benefits to our approach. On the STL-10 dataset with only 40 labels, InterLUDE achieves 3.2% error rate, while the best previous method reports 14.9%.