視覚言語行動モデルにおける線形表現仮説
The Linear Representation Hypothesis for Vision-Language-Action Models
視覚言語行動モデル（VLA）において、物理量の未来の変化を線形に読み出し・操作できる表現が存在することを理論的に示し、ナビゲーション実験で検証した。
詳しい要約
1. どんなもの？
2. 先行研究と比べてどこがすごい？
3. 技術・手法の肝は？
4. どうやって有効だと検証した？
5. 議論はある？
6. 次に読むべき論文は？
※ AIが要旨から生成した要約です。正確性は原文をご確認ください。
著者: Minseok Jeong, Hyewon Choi, Hiroyasu Tsukamoto, SooJean Han
分類: cs.LG, cs.AI
原文アブストラクト
The linear representation hypothesis (LRH) has become a standard lens for measuring and intervening on semantic information through the internal representations of large language models (LLMs). A growing body of work has begun extending this perspective to vision-language-action (VLA) models, but the dynamical nature of embodied interaction introduces an additional challenge. Unlike semantic attributes commonly studied in LLMs, such as gender or language, a physical quantity of interest (QoI) in a VLA evolves jointly with the system dynamics: the representation influences the actions selected by the policy, which alter the physical state and, in turn, the next representation. In this paper, we develop a theoretical, signature-based formulation of the LRH for VLA that unifies representations and policies. On the representation side, we establish the existence of representations from which the future evolution of a QoI under a candidate action trajectory can be recovered via linear probing. On the policy side, we introduce a signature generalized linear model for stochastic action chunks. This structure yields a monotonic change in the expected future QoI along linear paths in natural parameter space, enabling linear steering. We construct an explicit oracle representation in a planar control-affine navigation experiment and verify the predicted linear probing and steering mechanisms.