pNNCLR: Stochastic Pseudo Neighborhoods for Contrastive Learning based Unsupervised Representation Learning Problems

TL;DR

pNNCLR introduces stochastic pseudo nearest neighbors and a smooth-weight-update strategy to enhance contrastive learning, significantly improving support set quality and outperforming baseline methods in image recognition tasks.

Contribution

The paper proposes pNNCLR, a novel SSL approach using pseudo nearest neighbors and stochastic sampling to improve support set quality and learning stability.

Findings

Up to 8% performance improvement over NNCLR.

Comparable results to other SSL methods.

Effective on multiple image recognition datasets.

Abstract

Nearest neighbor (NN) sampling provides more semantic variations than pre-defined transformations for self-supervised learning (SSL) based image recognition problems. However, its performance is restricted by the quality of the support set, which holds positive samples for the contrastive loss. In this work, we show that the quality of the support set plays a crucial role in any nearest neighbor based method for SSL. We then provide a refined baseline (pNNCLR) to the nearest neighbor based SSL approach (NNCLR). To this end, we introduce pseudo nearest neighbors (pNN) to control the quality of the support set, wherein, rather than sampling the nearest neighbors, we sample in the vicinity of hard nearest neighbors by varying the magnitude of the resultant vector and employing a stochastic sampling strategy to improve the performance. Additionally, to stabilize the effects of uncertainty in NN-based learning, we employ a smooth-weight-update approach for training the proposed network. Evaluation of the proposed method on multiple public image recognition and medical image recognition datasets shows that it performs up to 8 percent better than the baseline nearest neighbor method, and is comparable to other previously proposed SSL methods.