🤖 AI Summary
Existing semi-supervised learning methods primarily enforce consistency across multiple views of individual samples, neglecting the structured relationships among unlabeled samples within a mini-batch.
Method: This paper proposes an intra-batch relational modeling framework that explicitly encodes the pairwise similarity structure among samples in the same batch as a relation matrix, and introduces Matrix Cross-Entropy (MCE) loss to align relational structures at the batch level—departing from conventional point-wise consistency paradigms. The method integrates contrastive relational modeling, consistency regularization, and strong-weak data augmentation.
Contribution/Results: On STL-10 with only 40 labeled samples, our method achieves 82.3% accuracy—outperforming FlexMatch by 15.21%. It consistently surpasses state-of-the-art methods including FixMatch and FlexMatch on multiple benchmarks such as CIFAR-10, CIFAR-100, and SVHN, demonstrating superior generalization and robustness in low-label regimes.
📝 Abstract
Semi-supervised learning has achieved notable success by leveraging very few labeled data and exploiting the wealth of information derived from unlabeled data. However, existing algorithms usually focus on aligning predictions on paired data points augmented from an identical source, and overlook the inter-point relationships within each batch. This paper introduces a novel method, RelationMatch, which exploits in-batch relationships with a matrix cross-entropy (MCE) loss function. Through the application of MCE, our proposed method consistently surpasses the performance of established state-of-the-art methods, such as FixMatch and FlexMatch, across a variety of vision datasets. Notably, we observed a substantial enhancement of 15.21% in accuracy over FlexMatch on the STL-10 dataset using only 40 labels. Moreover, we apply MCE to supervised learning scenarios, and observe consistent improvements as well.
