跳到正文
arXiv cs.AI· Xianyao Li, Ruitong Tian, Rui Min, Fang Xu, Jing Du·· 2 小时前AI 评分34

可执行感知记忆(EPM):从感知到想象的世界模型框架

Perception into Imagination: Executable Memory for World Models

AI 导读

研究者提出可执行感知记忆(EPM)框架,让世界模型在想象过程中维持显式感知状态,其演化读取信号用于奖励预测、价值估计和策略学习。在 Memory Maze 13×13 上,EPM 在 1M 交互下达到 30.37 回报,比仅 informs 策略与价值的基线 23.37 高出 7.00,并在 9×9 和 MiniWorld 中改善控制表现。

正文 · 原文

View PDF HTML (experimental)

Abstract:Physical agents must use what they previously saw or touched after visibility or contact is lost. Their actions change the implications of these remembered facts. We introduce executable perceptual memory (EPM), a world-model framework that advances explicit perceptual state alongside neural latent dynamics during imagination. Its evolving read-out informs reward prediction, value estimation and policy learning. We align reward inputs with the transitions that produced them and supervise the reward head on open-loop rollouts of recorded actions. On Memory Maze $13\times13$, EPM reaches 30.37 return at 1M interactions, compared with 23.37 when the same evolving read-out informs only policy and value on a common perceptual front end: a gain of 7.00 [3.80, 10.21]. The complete method also improves control on $9\times9$ and in MiniWorld. Probes of remembered, out-of-view arrivals show stronger reward anticipation in both visual implementations. Concentration, tactile recovery and real tactile frames demonstrate complementary roles of remembered facts in imagined-policy training, motion-dependent decisions and sensory-content retention. EPM connects persistent perception to the predicted consequences and learning signals of a world model for Physical AI.
Comments: 18 pages, 6 figures, 15 tables
Subjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computer Vision and Pattern Recognition (cs.CV); Robotics (cs.RO)
Cite as: arXiv:2611.22294 [cs.LG]
  (or arXiv:2611.22294v1 [cs.LG] for this version)
  https://doi.org/10.48550/arXiv.2611.22294

arXiv-issued DOI via DataCite (pending registration)

Submission history

From: Xianyao Li [view email]
[v1] Tue, 17 Nov 2026 06:52:02 UTC (390 KB)

来源:arXiv cs.AI · arxiv.org