Model State Arithmetic for Machine Unlearning

• URL: http://arxiv.org/abs/2506.20941v1
• Date: Thu, 26 Jun 2025 02:16:16 GMT
• Title: Model State Arithmetic for Machine Unlearning
• Authors: Keivan Rezaei, Mehrdad Saberi, Abhilasha Ravichander, Soheil Feizi,
• Abstract summary: We propose a new algorithm, MSA, for estimating and undoing the influence of datapoints.<n>Our experimental results demonstrate that MSA consistently outperforms existing machine unlearning algorithms.
• Score: 43.773053236733425
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Large language models are trained on massive corpora of web data, which may include private data, copyrighted material, factually inaccurate data, or data that degrades model performance. Eliminating the influence of such problematic datapoints through complete retraining -- by repeatedly pretraining the model on datasets that exclude these specific instances -- is computationally prohibitive. For this reason, unlearning algorithms have emerged that aim to eliminate the influence of particular datapoints, while otherwise preserving the model -- at a low computational cost. However, precisely estimating and undoing the influence of individual datapoints has proved to be challenging. In this work, we propose a new algorithm, MSA, for estimating and undoing the influence of datapoints -- by leveraging model checkpoints i.e. artifacts capturing model states at different stages of pretraining. Our experimental results demonstrate that MSA consistently outperforms existing machine unlearning algorithms across multiple benchmarks, models, and evaluation metrics, suggesting that MSA could be an effective approach towards more flexible large language models that are capable of data erasure.

Related papers

• Losing is for Cherishing: Data Valuation Based on Machine Unlearning and Shapley Value [18.858879113762917]
  We propose Unlearning Shapley, a novel framework that leverages machine unlearning to estimate data values efficiently.<n>Our method computes Shapley values via Monte Carlo sampling, avoiding retraining and eliminating dependence on full data.<n>This work bridges the gap between data valuation theory and practical deployment, offering a scalable, privacy-compliant solution for modern AI ecosystems.
  arXiv  Detail & Related papers  (2025-05-22T02:46:03Z)
• Capturing the Temporal Dependence of Training Data Influence [100.91355498124527]
  We formalize the concept of trajectory-specific leave-one-out influence, which quantifies the impact of removing a data point during training.<n>We propose data value embedding, a novel technique enabling efficient approximation of trajectory-specific LOO.<n>As data value embedding captures training data ordering, it offers valuable insights into model training dynamics.
  arXiv  Detail & Related papers  (2024-12-12T18:28:55Z)
• RESTOR: Knowledge Recovery in Machine Unlearning [71.75834077528305]
  Large language models trained on web-scale corpora can contain private or sensitive information.<n>Several machine unlearning algorithms have been proposed to eliminate the effect of such datapoints.<n>We propose the RESTOR framework for machine unlearning evaluation.
  arXiv  Detail & Related papers  (2024-10-31T20:54:35Z)
• Attribute-to-Delete: Machine Unlearning via Datamodel Matching [65.13151619119782]
  Machine unlearning -- efficiently removing a small "forget set" training data on a pre-divertrained machine learning model -- has recently attracted interest. Recent research shows that machine unlearning techniques do not hold up in such a challenging setting.
  arXiv  Detail & Related papers  (2024-10-30T17:20:10Z)
• Data Shapley in One Training Run [88.59484417202454]
  Data Shapley provides a principled framework for attributing data's contribution within machine learning contexts.<n>Existing approaches require re-training models on different data subsets, which is computationally intensive.<n>This paper introduces In-Run Data Shapley, which addresses these limitations by offering scalable data attribution for a target model of interest.
  arXiv  Detail & Related papers  (2024-06-16T17:09:24Z)
• Unlearning Traces the Influential Training Data of Language Models [31.33791825286853]
  This paper presents UnTrac: unlearning traces the influence of a training dataset on the model's performance. We propose a more scalable approach, UnTrac-Inv, which unlearns a test dataset and evaluates the unlearned model on training datasets.
  arXiv  Detail & Related papers  (2024-01-26T23:17:31Z)
• Machine Learning Based Missing Values Imputation in Categorical Datasets [2.5611256859404983]
  This research looked into the use of machine learning algorithms to fill in the gaps in categorical datasets. The emphasis was on ensemble models constructed using the Error Correction Output Codes framework. Deep learning for missing data imputation has obstacles despite these encouraging results, including the requirement for large amounts of labeled data.
  arXiv  Detail & Related papers  (2023-06-10T03:29:48Z)
• How Training Data Impacts Performance in Learning-based Control [67.7875109298865]
  This paper derives an analytical relationship between the density of the training data and the control performance. We formulate a quality measure for the data set, which we refer to as $rho$-gap. We show how the $rho$-gap can be applied to a feedback linearizing control law.
  arXiv  Detail & Related papers  (2020-05-25T12:13:49Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.