ProtoCon: Pseudo-label Refinement via Online Clustering and Prototypical Consistency for Efficient Semi-supervised Learning

TL;DR

ProtoCon is a semi-supervised learning method that refines pseudo-labels through online clustering and prototypical consistency, improving performance and convergence speed on multiple datasets.

Contribution

ProtoCon introduces a novel pseudo-label refinement technique using online clustering and prototypical loss, scalable to large datasets without storing embeddings.

Findings

Significant performance improvements over state-of-the-art methods.

Faster convergence on multiple datasets including CIFAR, ImageNet, and DomainNet.

Effective label refinement in label-scarce semi-supervised learning scenarios.

Abstract

Confidence-based pseudo-labeling is among the dominant approaches in semi-supervised learning (SSL). It relies on including high-confidence predictions made on unlabeled data as additional targets to train the model. We propose ProtoCon, a novel SSL method aimed at the less-explored label-scarce SSL where such methods usually underperform. ProtoCon refines the pseudo-labels by leveraging their nearest neighbours' information. The neighbours are identified as the training proceeds using an online clustering approach operating in an embedding space trained via a prototypical loss to encourage well-formed clusters. The online nature of ProtoCon allows it to utilise the label history of the entire dataset in one training cycle to refine labels in the following cycle without the need to store image embeddings. Hence, it can seamlessly scale to larger datasets at a low cost. Finally, ProtoCon addresses the poor training signal in the initial phase of training (due to fewer confident predictions) by introducing an auxiliary self-supervised loss. It delivers significant gains and faster convergence over state-of-the-art across 5 datasets, including CIFARs, ImageNet and DomainNet.