Rethinking Latent Redundancy in Behavior Cloning: An Information Bottleneck Approach for Robot Manipulation

TL;DR

This paper introduces an information-theoretic approach using the Information Bottleneck principle to reduce redundancy in latent representations, improving behavior cloning for robot manipulation.

Contribution

It is the first to apply the IB principle to behavior cloning, providing a theoretical framework to quantify and mitigate redundancy in learned representations.

Findings

Significant performance improvements on CortexBench and LIBERO benchmarks.

Redundancy reduction leads to better generalization in robot manipulation tasks.

Extensive analysis confirms the effectiveness of the IB approach in BC.

Abstract

Behavior Cloning (BC) is a widely adopted visual imitation learning method in robot manipulation. Current BC approaches often enhance generalization by leveraging large datasets and incorporating additional visual and textual modalities to capture more diverse information. However, these methods overlook whether the learned representations contain redundant information and lack a solid theoretical foundation to guide the learning process. To address these limitations, we adopt an information-theoretic perspective and introduce mutual information to quantify and mitigate redundancy in latent representations. Building on this, we incorporate the Information Bottleneck (IB) principle into BC, which extends the idea of reducing redundancy by providing a structured framework for compressing irrelevant information while preserving task-relevant features. This work presents the first comprehensive study on redundancy in latent representations across various methods, backbones, and experimental settings, while extending the generalizability of the IB to BC. Extensive experiments and analyses on the CortexBench and LIBERO benchmarks demonstrate significant performance improvements with IB, underscoring the importance of reducing input data redundancy and highlighting its practical value for more practical applications. Project Page: https://baishuanghao.github.io/BC-IB.github.io.