新研究将持续学习的范畴从遗忘和可塑性扩展 · 追踪5个来源

报道来源 [7]

1. arXiv cs.AI TIER_1 English(EN) · Timm Hess, Abhishek Jha, Gido M. van de Ven, Tinne Tuytelaars · 2026-08-20 04:00

   遗忘、可塑性与共同观察：持续学习的第三个层面

   arXiv:2608.18803v1 Announce Type: cross Abstract: Efficient continual learning remains a fundamental challenge for deep neural networks. While catastrophic forgetting and loss of plasticity are widely considered the primary obstacles to overcome, we show that these two issues can…

2. arXiv cs.AI TIER_1 English(EN) · Borui Kang, Jinrui Gu, Junhan Lv, Wenbin Li, Lei Wang, Yang Gao · 2026-08-20 04:00

   利用持续学习：超越模型参数的持续适应

   arXiv:2608.19013v1 Announce Type: cross Abstract: Continual learning has largely been model-centric, treating model parameters as the state that changes with sequential experience. Modern agents can also adapt through a harness of prompts, memories, tools, skills, and routing rul…

3. Hugging Face Daily Papers TIER_1 English(EN) · 2026-08-19 15:12

   利用持续学习：超越模型参数的持续适应

   Continual learning has largely been model-centric, treating model parameters as the state that changes with sequential experience. Modern agents can also adapt through a harness of prompts, memories, tools, skills, and routing rules. Because these contents jointly shape later exe…

4. arXiv cs.LG TIER_1 English(EN) · Maksim A. Kazanskii · 2026-08-18 04:00

   遗忘的几何学：持续学习中的表征流变

   arXiv:2608.15854v1 Announce Type: new Abstract: Catastrophic forgetting remains a fundamental obstacle to continual learning, where neural networks lose previously acquired knowledge while learning new tasks. Existing methods primarily mitigate forgetting through parameter regula…

5. arXiv cs.LG TIER_1 English(EN) · Zhiming Xu, Huiyu Yi, Zhen-Hao Xie, Baile Xu, Furao Shen, Jian Zhao, Suorong Yang · 2026-08-18 04:00

   面向预训练模型式持续学习的任务锚定表征塑造

   arXiv:2608.16345v1 Announce Type: new Abstract: Pre-trained models (PTMs) provide a strong foundation for continual learning by offering stable representations that facilitate lightweight adaptation to new tasks. However, adapting well to each task does not ensure reliable infere…

6. arXiv cs.CV TIER_1 English(EN) · Jiaqi Wang, Zhou Fang, Qiongfeng Shi, Yi Zhou · 2026-08-21 04:00

   OrthoSkillVLA：通过梯度感知技能子空间自适应实现持续技能学习

   arXiv:2608.19589v1 Announce Type: cross Abstract: Pretrained Vision-Language-Action models provide a strong foundation for robot learning, but sequentially adapting them to diverse skills can perturb the representations and velocity mappings used by previous skills, leading to ca…

7. X — Omar Sanseviero (HF research) TIER_1 English(EN) · omarsar0 · 2026-08-20 20:17

   关于持续学习的精彩论文。

   Banger paper on harness continual learning. (bookmark it) If you already are allowing your agents to rewrite their own prompts, skills, or memory files, this one is worth your time. (bookmark it) Continual learning has always tracked what changes in the weights. Modern agents…