跳到正文
arXiv cs.AI· Nishit Anand, Ramani Duraiswami, Dinesh Manocha·· 4 小时前AI 评分41

世界模型应该遗忘什么?面向持续适应的分层保留策略

What Should World Models Forget? Stratified Retention for Continual Adaptation

AI 导读

一篇 NeurIPS 2026 持续世界模型研讨会论文提出,持续学习的世界模型应按不变性时间尺度分层保留知识:物理定律等不变量只能精化不可退化,环境变化后的实例级事实则应被修订。作者指出标准遗忘指标会把正确修订误判为遗忘,从而把冻结模型排为最优,并提出差分保留方法,分别报告不变量保留率和修订延迟两项指标。

正文 · 原文

View PDF HTML (experimental)

Abstract:Continual learning treats any drop in performance on previously seen data as forgetting, a convention inherited from settings where the prediction target is stationary. World models violate this assumption: their target is the environment, which changes, so knowledge that was accurate when learned can later become false, and revising it is correct behavior rather than a defect. Non-stationary ground truth is well studied in the concept drift literature and in the temporal factuality of language models, but it has not been formulated for world models, which are distinctive in that they also encode structural knowledge, such as physical laws and object permanence, that must never be degraded. We argue that continual world models need retention stratified by invariance timescale, the time over which knowledge stays true: invariants should be refined but never degraded, while instance-level facts should be revised once the environment changes. We show that standard forgetting metrics, used alone, count correct revision as forgetting and so rank a frozen model highest. We propose differential retention, which reports two quantities separately: invariant retention, checked after every adaptation round, and revision latency, the number of observations a model needs to pick up a change.
Comments: Accepted to NeurIPS 2026 Workshop on Continual World Models. Project Website: this https URL
Subjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computer Vision and Pattern Recognition (cs.CV); Image and Video Processing (eess.IV); Signal Processing (eess.SP)
Cite as: arXiv:2610.03713 [cs.LG]
  (or arXiv:2610.03713v2 [cs.LG] for this version)
  https://doi.org/10.48550/arXiv.2610.03713

arXiv-issued DOI via DataCite

Submission history

From: Nishit Anand [view email]
[v1] Fri, 2 Oct 2026 17:58:14 UTC (255 KB)
[v2] Thu, 12 Nov 2026 21:35:23 UTC (242 KB)

来源:arXiv cs.AI · arxiv.org