analiza aiqbench
This research introduces CE$^4$L, a novel benchmark addressing the gap in continual learning for embodied agents by simultaneously handling task shifts and viewpoint changes across ego and exo cameras. The authors propose VISTA, a parameter-efficient method using lightweight adapters and training-free routing that achieves state-of-the-art performance on these complex multi-view tasks. This work is significant for advancing robust perception systems in real-world scenarios where camera perspectives and objectives evolve continuously.
CE$^4$L: Continual Ego, Exo, and Ego-Exo Learning
arXiv:2609.23492v1 Announce Type: cross Abstract: Perception for embodied agents is video-based, often multi-view (ego, exo, or both), and inherently continual, with simultaneous task and viewpoint shifts. Yet continual learning (CL) remains dominated by exo-only recognition tasks, obscuring behavior under these real-world coupled shifts. We introduce Continual Ego, E}xo, and Ego-Exo Learning (CE$^4$L), a unified multi-view CL benchmark spanning four representative tasks: cross-view referenced skill assessment, temporal action segmentation, cross-view association, and action anticipation & planning. CE$^4$L highlights challenges largely absent in prior CL benchmarks, including cross-view correspondence, view-dependent asynchrony, and heterogeneous semantic objectives. To this end, we propose Video Incremental Subspace-routed Task Adapters (VISTA), a parameter-efficient baseline method that stores task-specific updates in lightweight adapters and performs training-free routing via residual distance to task-specific whitened subspaces estimated from second-order statistics. Extensive experiments demonstrate the significantly varied efficacy of representative CL methods across CE$^4$L settings, while VISTA is consistently competitive and achieves state-of-the-art overall performance. Our source code for benchmarks and methods is available at https://github.com/AnAppleCore/CE4L .
Abstract:Perception for embodied agents is video-based, often multi-view (ego, exo, or both), and inherently continual, with simultaneous task and viewpoint shifts. Yet continual learning (CL) remains dominated by exo-only recognition tasks, obscuring behavior under these real-world coupled shifts. We introduce Continual Ego, E}xo, and Ego-Exo Learning (CE$^4$L), a unified multi-view CL benchmark spanning four representative tasks: cross-view referenced skill assessment, temporal action segmentation, cross-view association, and action anticipation & planning. CE$^4$L highlights challenges largely absent in prior CL benchmarks, including cross-view correspondence, view-dependent asynchrony, and heterogeneous semantic objectives. To this end, we propose Video Incremental Subspace-routed Task Adapters (VISTA), a parameter-efficient baseline method that stores task-specific updates in lightweight adapters and performs training-free routing via residual distance to task-specific whitened subspaces estimated from second-order statistics. Extensive experiments demonstrate the significantly varied efficacy of representative CL methods across CE$^4$L settings, while VISTA is consistently competitive and achieves state-of-the-art overall performance. Our source code for benchmarks and methods is available at this https URL .
23 pages. Accepted by ICML 2026
Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
arXiv:2609.23492 [cs.CV]
(or arXiv:2609.23492v1 [cs.CV] for this version)
https://doi.org/10.48550/arXiv.2609.23492 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Text integral sub licența deschisă a publicației · 269 cuvinte
Subiecte conexe
What's Continual Learning, and Why Might We Expect To See It In Advanced LLM Agents?
LessWrong · 12 iun. 2026
Benchtalks #3: We taught AI everything except how to learn
Snorkel AI Blog · 25 iun. 2026
Continual learning might make your blocking monitors nearly useless
AI Alignment Forum · 24 sept. 2026
Continual Learning Bench: measuring whether AI systems actually improve with experience
Snorkel AI Blog · 21 aug. 2026