← 研踪 PaperTrailarXiv:2610.10857 · v2
Self-Supervised Keyframe Discovery for Horizon-Invariant Behavior Cloning
首次提交 2026-10-08 04:01
版本更新 2026-11-07 14:06
雷达观测 2026-11-10 11:14
摘要
Behavior cloning (BC) in non-Markovian environments is a challenging problem because policies have to reason over contextual information over long horizons. Existing policy architectures rely on recurrent or attention-based mechanisms to capture long-term dependencies. However, recurrent models suffer from hidden-state collapse and gradient instability under backpropagation through time while attention-based models are fundamentally limited by context length. To address these issues, we propose Keyframe Mnemonics, a novel self-supervised method that a set of information-critical observations () by learning an objective from randomly sampled past observations and using it as a reward for keyframe selection. We then train a BC policy that conditions on the discovered keyframes to model the action distribution. Under the simplicity bias assumption, our formulation provides context retention guarantees over an infinite horizon, while maintaining a small set of decision-relevant keyframes in the policy's working memory. We evaluate our method on synthetic memory domains, where mnemonic-conditioned BC policies achieve % success rates (SR) and generalize to horizons orders of magnitude beyond training without performance degradation. Moreover, we evaluate on a memory-intensive robot manipulation benchmark, achieving a % average absolute SR improvement over the strongest baseline across tasks. Finally, we demonstrate generalization with % SR at longer time horizons on a real robot. Code and videos available at https://keyframe-mnemonics.github.io.
作者备注:Accepted at NeurIPS 2026
观测到的版本
相邻记录可能跳过中间版本;首次发现 v2 及以上也会标为版本更新。
v2 · 2026-11-07 14:06 · 最新观测版本
Self-Supervised Keyframe Discovery for Horizon-Invariant Behavior Cloning
Behavior cloning (BC) in non-Markovian environments is a challenging problem because policies have to reason over contextual information over long horizons. Existing policy architectures rely on recurrent or attention-based mechanisms to capture long-term dependencies. However, recurrent models suffer from hidden-state collapse and gradient instability under backpropagation through time while attention-based models are fundamentally limited by context length. To address these issues, we propose Keyframe Mnemonics, a novel self-supervised method that a set of information-critical observations () by learning an objective from randomly sampled past observations and using it as a reward for keyframe selection. We then train a BC policy that conditions on the discovered keyframes to model the action distribution. Under the simplicity bias assumption, our formulation provides context retention guarantees over an infinite horizon, while maintaining a small set of decision-relevant keyframes in the policy's working memory. We evaluate our method on synthetic memory domains, where mnemonic-conditioned BC policies achieve % success rates (SR) and generalize to horizons orders of magnitude beyond training without performance degradation. Moreover, we evaluate on a memory-intensive robot manipulation benchmark, achieving a % average absolute SR improvement over the strongest baseline across tasks. Finally, we demonstrate generalization with % SR at longer time horizons on a real robot. Code and videos available at https://keyframe-mnemonics.github.io.
查看此版本解读 →查看此版本原文 ↗