DreamSAC: 対称性探索によるハミルトン世界モデルの学習
DreamSAC: Learning Hamiltonian World Models via Symmetry Exploration
物理法則の不変性を能動的に探索する「対称性探索」と、ハミルトン力学に基づく世界モデルを組み合わせ、外挿一般化を実現するフレームワークを提案した。
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: Jinzhou Tang, Fan Feng, Minghao Fu, Wenjun Lin, Biwei Huang, Keze Wang
分類: cs.CV, cs.AI, cs.LG
原文アブストラクト
Learned world models excel at interpolative generalization but fail at extrapolative generalization to novel physical properties. This limitation arises because they learn statistical correlations rather than the environment's underlying generative rules, such as physical invariances and conservation laws. We argue that learning these invariances is key to robust extrapolation. To achieve this, we first introduce \textbf{Symmetry Exploration}, an unsupervised exploration strategy where an agent is intrinsically motivated by a Hamiltonian-based curiosity bonus to actively probe and challenge its understanding of conservation laws, thereby collecting physically informative data. Second, we design a Hamiltonian-based world model that learns from the collected data, using a novel self-supervised contrastive objective to identify the invariant physical state from raw, view-dependent pixel observations. Our framework, \textbf{DreamSAC}, trained on this actively curated data, significantly outperforms state-of-the-art baselines in 3D physics simulations on tasks requiring extrapolation.
