Continual Learning in Transition

Source

Source name: arXiv
Source kind: paper
Authors / organization: Zhiyan Hou, Dan Zhang, Tao Feng, Liyuan Wang, Wei Li, Xiangzhao Hao, Hongyan An, Junfeng Fang, Haokai Ma, Zhaohui Xu, Haiyun Guo, Jinqiao Wang, Tat-Seng Chua
Venue: arXiv: cs.LG
Published: 2026-08-06
Collected: 2026-08-07
URL: http://arxiv.org/abs/2608.06216v1
PDF: https://arxiv.org/pdf/2608.06216v1
DOI: UNKNOWN

Plain-Language Brief

The text discusses a shift in continual learning, moving away from traditional parameter-centric methods towards system-level adaptation. The authors propose a tri-axial framework to analyze this transition: When, How, and Where learning occurs. They survey various methods, discuss the implications of this shift, and outline future directions.

What the Document Says

The text is about continual learning and its transition from parameter-centric methods towards system-level adaptation.
The authors use a tri-axial framework to examine this transition: When, How, and Where learning occurs.
Off-policy, on-policy, and beyond-gradient optimization mechanics fall under the How dimension.
Pre-training, post-training, and inference-time stages are included in the When dimension.
Updates within internal parameters versus external structural constraints are part of the Where dimension.
The text discusses representative methods, the ongoing transition of continual learning, key challenges, broader implications, and future directions.

Key Concepts Mentioned

continual learning
parameter-centric methods
system-level adaptation
tri-axial framework
off-policy
on-policy
beyond-gradient optimization
pre-training
post-training
inference-time
internal parameters
external structural constraints

Claims Made by the Document

There is a transition from parameter-centric learning towards system-level adaptation in continual learning.
The authors propose a tri-axial framework to analyze this transition.

Evidence Used

The text discusses the evolution of continual learning towards system-level adaptation.
The authors present a tri-axial framework consisting of off-policy, on-policy, and beyond-gradient optimization mechanics, pre-training, post-training, inference-time stages, and internal parameters versus external structural constraints.

Methods / Approach

UNKNOWN

Limitations Stated by the Authors

UNKNOWN

Limitations Noted by the Agent

Agent interpretation — not from the source.

UNKNOWN

Open Questions

UNKNOWN

Notes for Future Q&A

May include interpretation. Labeled where it does.

UNKNOWN

Raw Abstract / Excerpt

Classical continual learning (CL) has primarily focused on enabling models to update and retain knowledge through parameter-centric mechanisms, e.g., training strategies, architectural designs, and weight adaptation. However, emerging paradigms are reshaping the scope of CL beyond this traditional model adaptation view. For instance, on-policy learning broadens the space of update mechanisms; test-time training extends CL from the training phase to inference; and external harness components such as memory, skill libraries, and interaction protocols extend the evolutionary boundaries of model capabilities far beyond the static parameter space. Collectively, these developments indicate a transition from parameter-centric learning toward system-level adaptation. To characterize this transition, we examine the evolution of continual learning through three dimensions: When, How, and Where learning occurs. The How dimension encompasses off-policy, on-policy, and beyond-gradient optimization mechanics. The When dimension captures evolution across pre-training, post-training, and inference-time stages. The Where dimension delineates updates occurring within internal parameters versus external structural constraints. Anchored by this tri-axial framework, we systematically survey representative methods, trace the ongoing transition of continual learning, and discuss the key challenges, broader implications, and future directions arising from this paradigm shift.