Masked Reconstruction Contrastive Learning with Information Bottleneck Principle

TL;DR

This paper introduces a novel contrastive learning framework based on the Information Bottleneck principle, using masked reconstruction to enhance generalization and reduce overfitting to discriminative features, with superior results across multiple vision tasks.

Contribution

It proposes the Masked Reconstruction Contrastive Learning (MRCL) model that integrates IB principles, combining masking and reconstruction to improve generalization in contrastive learning.

Findings

MRCL outperforms existing models on image classification, semantic segmentation, and object detection.

Theoretical analysis confirms optimal information compression and representation.

Masked augmentation effectively eliminates redundant and noisy information.

Abstract

Contrastive learning (CL) has shown great power in self-supervised learning due to its ability to capture insight correlations among large-scale data. Current CL models are biased to learn only the ability to discriminate positive and negative pairs due to the discriminative task setting. However, this bias would lead to ignoring its sufficiency for other downstream tasks, which we call the discriminative information overfitting problem. In this paper, we propose to tackle the above problems from the aspect of the Information Bottleneck (IB) principle, further pushing forward the frontier of CL. Specifically, we present a new perspective that CL is an instantiation of the IB principle, including information compression and expression. We theoretically analyze the optimal information situation and demonstrate that minimum sufficient augmentation and information-generalized representation are the optimal requirements for achieving maximum compression and generalizability to downstream tasks. Therefore, we propose the Masked Reconstruction Contrastive Learning~(MRCL) model to improve CL models. For implementation in practice, MRCL utilizes the masking operation for stronger augmentation, further eliminating redundant and noisy information. In order to alleviate the discriminative information overfitting problem effectively, we employ the reconstruction task to regularize the discriminative task. We conduct comprehensive experiments and show the superiority of the proposed model on multiple tasks, including image classification, semantic segmentation and objective detection.