ATLAS: Aligned Transport of Latent Structure for Reliable World Model Planning
- 类型：arxiv
- 标识：2609.36333
- 链接：https://arxiv.org/abs/2609.36333
- 主分类：engineering
- 形态：method
- TLDR：Latent world models rely on representation geometry for planning, yet regularizing the latent marginal alone does not determine the state-to-state relationships used for action selection. We show that this can cause planning-relevant novelty structure to be weakened as representations are transformed into the final latent used by the planner. We introduce Aligned Transport of Latent Structure (ATLAS), a training objective that explicitly preserves relational geometry while calibrating the global latent distribution. ATLAS transfers normalized pairwise structure from an informative encoder repr
- 待LLM分类：否
- 标题中文：ATLAS：面向可靠世界模型规划的对齐潜在结构传输
- TLDR中文：Latent world models 依赖表征几何进行规划，但仅正则化 latent marginal 并不能决定用于动作选择的 state-to-state 关系。我们表明，当表征被转换为 planner 所使用的最终 latent 时，规划相关的新颖性结构会被削弱。我们提出 Aligned Transport of Latent Structure (ATLAS)，一种在标定全局 latent 分布的同时显式保持关系几何的训练目标。ATLAS 将归一化的成对结构从一个 informative encoder 的表征中迁移过来
- 来源文件：
- /inbox/tom/_candidates/2026-10-01-agent-rag-longcontext-candidates.json