Self-supervised Representation Learning With Relative Predictive Coding

Abstract

This paper introduces Relative Predictive Coding (RPC), a new contrastive representation learning objective that maintains a good balance among training stability, minibatch size sensitivity, and downstream task performance. The key to the success of RPC is two-fold. First, RPC introduces the relative parameters to regularize the objective for boundedness and low variance. Second, RPC contains no logarithm and exponential score functions, which are the main cause of training instability in prior contrastive objectives. We empirically verify the effectiveness of RPC on benchmark vision and speech self-supervised learning tasks. Lastly, we relate RPC with mutual information (MI) estimation, showing RPC can be used to estimate MI with low variance.

Stats

• citations0
• S2 citations—
• github stars0
• HF likes0
• heat score0.00
• arxiv keytsai2021self

Related papers

• Non-autoregressive Predictive Coding For Learning Speech Representations From Local Dependencies (2020)12.47
• Aligned Contrastive Predictive Coding (2021)9.23
• Contrastive Separative Coding For Self-supervised Representation Learning (2021)0.00
• Guided Contrastive Self-supervised Pre-training For Automatic Speech Recognition (2022)0.00
• Unsupervised Speech Segmentation And Variable Rate Representation Learning Using Segmental Contrastive Predictive Coding (2021)9.92
• Improved Speech Representations With Multi-target Autoregressive Predictive Coding (2020)10.97
• Learning Speech Representations With Variational Predictive Coding (2025)0.00
• Generative Pre-training For Speech With Autoregressive Predictive Coding (2019)14.73