PulseAugur
实时 07:40:30
English(EN) Why Zeroth-Order Adaptation May Forget Less: A Randomized Shaping Theory

新理论解释了为什么零阶适应能减少模型遗忘

研究人员开发了一个新的理论框架——随机塑造理论(Randomized Shaping Theory),来解释在持续学习中,零阶(ZO)适应方法为何可能比一阶(FO)方法遗忘更少。该理论认为,经过恰当分析的ZO适应,可以通过选择性地收缩适应的各向异性分量来保留更多先前获得的知识。这一理论洞见催生了一个名为RISE的新算法,它将校准过的ZO塑造应用于参数块内的精确FO梯度,以改善持续学习中的稳定-塑性权衡。 AI

影响 为持续学习的改进提供了理论解释,可能带来更强大的、能够随时间保留知识的AI系统。

排序理由 该集群包含一篇详细介绍持续学习的理论框架和提出的算法的新学术论文。[lever_c_demoted from research: ic=1 ai=1.0]

在 arXiv cs.LG 阅读 →

AI 生成摘要 · Google Gemini · 来自 1 个来源。 我们如何撰写摘要 →

新理论解释了为什么零阶适应能减少模型遗忘

本文如何被排名

Signal score
0 / 100
Composite score across the factors below. Higher = stronger signal that this story matters right now.
Newsworthiness bucket
Tool
该集群包含一篇详细介绍持续学习的理论框架和提出的算法的新学术论文。[lever_c_demoted from research: ic=1 ai=1.0]
Source corroboration
Single-source cluster
Only one publisher covered this so far. Single-source stories can still rank when the publisher is high-authority, but they lack cross-source corroboration.
Topics
paper, safety
Editorial topic classification. Feeds into how the story surfaces on /topic/<slug> hub pages and into the per-entity coverage mix.
AI-industry relevance
High
Clearly on-topic for AI-industry coverage.
Story freshness
113 days old
Aged out of breaking-news scoring windows; ranking reflects the durable signal from the full source set.

完整方法见我们的编辑标准

报道来源 [1]

  1. arXiv cs.LG TIER_1 English(EN) · Zhongxiang Dai ·

    为什么零阶适应可能遗忘更少:一种随机塑形理论

    Continual learning requires new-task adaptation without damaging previously acquired capabilities. Recent forward-pass and zeroth-order (ZO) results show that low-query adaptation may retain better than first-order (FO) descent, but the usual view of ZO as noisy FO estimation doe…