Lune

ICML2025Top-tier venue

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

Shuanghao Bai, Wanqi Zhou, Pengxiang Ding, Wei Zhao, Donglin Wang, Badong Chen

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 show consistent performance improvements with IB across various settings, underscoring the importance of reducing input data redundancy and highlighting its practical value for real-world applications. Project

Ask Lune

Know this paper in minutes, not an afternoon.

Lune can read every page of it and the top-tier papers that cite it, and gives you its claims, its evidence and its limits, each one cited to the paper it comes from. No skimming, no AI slop.

5 top-tier papers have cited it since. Know what they found before your reviewers do.

Questions to start from

Lune can draw on

the full paper · the top-tier papers that cite it · related work

Get the answer

Free to start. No credit card required.

Lineage

  • LatentVLA: Taming Latent Space for Generalizable and Long-Horizon Bimanual Manipulation cites Vision-Language Foundation Models as Effective Robot Imitators
  • ZPressor: Bottleneck-Aware Compression for Scalable Feed-Forward 3DGS cites An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale
  • ZPressor: Bottleneck-Aware Compression for Scalable Feed-Forward 3DGS cites Learning Robust Representations via Multi-View Information Bottleneck
  • Where are we in the search for an Artificial Visual Cortex for Embodied Intelligence? cites An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale
  • Where are we in the search for an Artificial Visual Cortex for Embodied Intelligence? cites Ego4D: Around the World in 3, 000 Hours of Egocentric Video
Builds onCited byLeft to right by year; the more cited, the higher.

Cited by top-tier papers5

  1. ZPressor: Bottleneck-Aware Compression for Scalable Feed-Forward 3DGSWeijie Wang, Donny Y. Chen, Zeyu Zhang, Duochao Shi et al.NeurIPS 2025 · 31 citations
  2. Latent Reasoning VLA: Latent Thinking and Prediction for Vision-Language-Action ModelsShuanghao Bai, Jing Lyu, Wanqi Zhou, Zhe Li et al.ICML 2026 · 15 citations
  3. A Comprehensive Information-Decomposition Analysis of Large Vision-Language ModelsLixin Xiu, Xufang Luo, Hideki NakayamaICLR 2026 · 4 citations
  4. LatentVLA: Taming Latent Space for Generalizable and Long-Horizon Bimanual ManipulationJunming WangAAAI 2026 · 1 citation
  5. Blindfolded Experts Generalize Better: Insights from Robotic Manipulation and VideogamesEv Zisselman, Mirco Mutti, Shelly Francis-Meretzki, Elisei Shafer et al.NeurIPS 2025 · 1 citation

Builds on23

  1. An Image is Worth 16x16 Words: Transformers for Image Recognition at ScaleAlexey Dosovitskiy, Lucas Beyer, Alexander Kolesnikov, Dirk Weissenborn et al.ICLR 2021 · 21,477 citations
  2. Ego4D: Around the World in 3, 000 Hours of Egocentric VideoKristen Grauman, Andrew Westbury, Eugene Byrne, Zachary Chavis et al.CVPR 2022 · 525 citations
  3. Vision-Language Foundation Models as Effective Robot ImitatorsXinghang Li, Minghuan Liu, Hanbo Zhang, Cunjun Yu et al.ICLR 2024 · 375 citations
  4. Where are we in the search for an Artificial Visual Cortex for Embodied Intelligence?Arjun Majumdar, Karmesh Yadav, Sergio Arnaud, Yecheng Jason Ma et al.NeurIPS 2023 · 336 citations
  5. Learning Robust Representations via Multi-View Information BottleneckMarco Federici, Anjan Dutta, Patrick Forré, Nate Kushman et al.ICLR 2020 · 330 citations

Related papers

  1. Information Retention via Learning Supplemental FeaturesZhipeng Xie, Yahe LiICLR 2024 · 1 citation
  2. Representation Learning with Conditional Information Flow MaximizationDou Hu, Lingwei Wei, Wei Zhou, Songlin HuACL 2024
  3. Drop-Bottleneck: Learning Discrete Compressed Representation for Noise-Robust ExplorationJaekyeom Kim, Minjung Kim, Dongyeon Woo, Gunhee KimICLR 2021 · 20 citations
  4. FlowNIB: An Information Bottleneck Analysis of Bidirectional vs. Unidirectional Language ModelsMd Kowsher, Nusrat Jahan Prottasha, Shiyun Xu, Shetu Mohanto et al.ICLR 2026
  5. Structured IB: Improving Information Bottleneck with Structured Feature LearningHanzhe Yang, Youlong Wu, Dingzhu Wen, Yong Zhou et al.AAAI 2025 · 6 citations
Dusk over the sea between two cliffs drawn in fine vertical lines