arXiv：cs.LG· Hammad Rizwan, Mahtab Sarvmaili, Hassan Sajjad, Ga Wu·· 13 小时前
机器遗忘的实例级难度：一个被忽视的视角
Instance-Level Difficulty: A Missing Perspective in Machine Unlearning
AI 导读
研究指出当前深度机器遗忘研究普遍只关注方法整体效果，忽视了单个训练样本的遗忘难度差异。作者通过多种遗忘算法与数据集的实例级分析，总结出四个使数据点难以被遗忘的因素，并实证表明这些因素与具体遗忘算法无关，只与目标模型及其训练数据相关。
正文
Abstract:Current research on deep machine unlearning primarily focuses on improving or evaluating the overall effectiveness of unlearning methods while overlooking the varying difficulty of unlearning individual training samples. As a result, the broader feasibility of machine unlearning remains under-explored. This paper studies the cruxes that make machine unlearning difficult through a thorough instance-level unlearning performance analysis over various unlearning algorithms and datasets. In particular, we summarize four factors that make unlearning a data point difficult, and we empirically show that these factors are independent of a specific unlearning algorithm but only relevant to the target model and its training data. Given these findings, we argue that machine unlearning research should pay attention to the instance-level difficulty of unlearning.
| Subjects: | Machine Learning (cs.LG) |
| Cite as: | arXiv:2410.03043 [cs.LG] |
| (or arXiv:2410.03043v3 [cs.LG] for this version) | |
| https://doi.org/10.48550/arXiv.2410.03043 arXiv-issued DOI via DataCite |
Submission history
From: Mahtab Sarvmaili [view email]
[v1]
Thu, 3 Oct 2024 23:41:42 UTC (4,946 KB)
[v2]
Fri, 21 Feb 2025 04:44:56 UTC (30,473 KB)
[v3]
Fri, 9 Oct 2026 00:07:47 UTC (30,278 KB)
来源：arXiv：cs.LG · arxiv.org