Saved in:
Bibliographic Details
Main Authors: Xu, Mengzhu, Liu, Hanzhi, Peng, Ningkang, Chen, Qianyu, Xiao, Canran
Format: Preprint
Published: 2025
Subjects:
Online Access:https://arxiv.org/abs/2512.00694
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866917114290372608
author Xu, Mengzhu
Liu, Hanzhi
Peng, Ningkang
Chen, Qianyu
Xiao, Canran
author_facet Xu, Mengzhu
Liu, Hanzhi
Peng, Ningkang
Chen, Qianyu
Xiao, Canran
contents Continual learning for video--language understanding is increasingly important as models face non-stationary data, domains, and query styles, yet prevailing solutions blur what should stay stable versus what should adapt, rely on static routing/capacity, or require replaying past videos. We aim to explicitly specify where stability lives and where plasticity should be focused under realistic memory and privacy constraints. We introduce Affordance-First Decomposition (AFD): videos are mapped to slowly varying affordance tokens that form a shared, time-aligned substrate, while a lightweight, query-routed, conflict-aware scheduler concentrates adaptation and grows capacity only when needed. The substrate is stabilized via weak alignment and teacher consistency, and training uses question-only replay. AFD achieves state-of-the-art across protocols: 51.6% average accuracy with -1.8% forgetting on domain-incremental VideoQA, ViLCo R@1@0.5 of 29.6% (MQ) and 20.7% (NLQ) with 18.4% stAP@0.25 (VQ), and 39.5% accuracy with -1.6% forgetting on time-incremental iVQA. Overall, AFD offers an explicit, interpretable split between a stable interaction-centered substrate and targeted adaptation.
format Preprint
id arxiv_https___arxiv_org_abs_2512_00694
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Affordance-First Decomposition for Continual Learning in Video-Language Understanding
Xu, Mengzhu
Liu, Hanzhi
Peng, Ningkang
Chen, Qianyu
Xiao, Canran
Computer Vision and Pattern Recognition
Continual learning for video--language understanding is increasingly important as models face non-stationary data, domains, and query styles, yet prevailing solutions blur what should stay stable versus what should adapt, rely on static routing/capacity, or require replaying past videos. We aim to explicitly specify where stability lives and where plasticity should be focused under realistic memory and privacy constraints. We introduce Affordance-First Decomposition (AFD): videos are mapped to slowly varying affordance tokens that form a shared, time-aligned substrate, while a lightweight, query-routed, conflict-aware scheduler concentrates adaptation and grows capacity only when needed. The substrate is stabilized via weak alignment and teacher consistency, and training uses question-only replay. AFD achieves state-of-the-art across protocols: 51.6% average accuracy with -1.8% forgetting on domain-incremental VideoQA, ViLCo R@1@0.5 of 29.6% (MQ) and 20.7% (NLQ) with 18.4% stAP@0.25 (VQ), and 39.5% accuracy with -1.6% forgetting on time-incremental iVQA. Overall, AFD offers an explicit, interpretable split between a stable interaction-centered substrate and targeted adaptation.
title Affordance-First Decomposition for Continual Learning in Video-Language Understanding
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2512.00694