LLM Unlearning via Neural Activation Redirection

• URL: http://arxiv.org/abs/2502.07218v2
• Date: Tue, 07 Oct 2025 21:36:07 GMT
• Title: LLM Unlearning via Neural Activation Redirection
• Authors: William F. Shen, Xinchi Qiu, Meghdad Kurmanji, Alex Iacob, Lorenzo Sani, Yihong Chen, Nicola Cancedda, Nicholas D. Lane,
• Abstract summary: We propose LUNAR, a novel unlearning method grounded in the Linear Representation Hypothesis.<n>We show that LUNAR achieves state-of-the-art unlearning performance and superior controllability.
• Score: 24.157334866277534
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: The ability to selectively remove knowledge from LLMs is highly desirable. However, existing methods often struggle with balancing unlearning efficacy and retain model utility, and lack controllability at inference time to emulate base model behavior as if it had never seen the unlearned data. In this paper, we propose LUNAR, a novel unlearning method grounded in the Linear Representation Hypothesis and operates by redirecting the representations of unlearned data to activation regions that expresses its inability to answer. We show that contrastive features are not a prerequisite for effective activation redirection, and LUNAR achieves state-of-the-art unlearning performance and superior controllability. Specifically, LUNAR achieves between 2.9x and 11.7x improvement in the combined unlearning efficacy and model utility score (Deviation Score) across various base models and generates coherent, contextually appropriate responses post-unlearning. Moreover, LUNAR effectively reduces parameter updates to a single down-projection matrix, a novel design that significantly enhances efficiency by 20x and robustness. Finally, we demonstrate that LUNAR is robust to white-box adversarial attacks and versatile in real-world scenarios, including handling sequential unlearning requests.

Related papers

• LLM Unlearning Under the Microscope: A Full-Stack View on Methods and Metrics [10.638045151201084]
  We present a principled taxonomy of twelve recent stateful unlearning methods.<n>We revisit the evaluation of unlearning effectiveness (UE), utility retention (UT), and robustness (Rob)<n>Our analysis shows that current evaluations, dominated by multiple-choice question (MCQ) accuracy, offer only a narrow perspective.
  arXiv  Detail & Related papers  (2025-10-08T23:47:05Z)
• Reliable Unlearning Harmful Information in LLMs with Metamorphosis Representation Projection [17.369869625390894]
  We propose a Metamorphosis Representation Projection (MRP) approach to machine unlearning.<n>By implementing projective transformations in the hidden state space of specific network layers, our method effectively eliminates harmful information while preserving useful knowledge.<n> Experimental results demonstrate that our approach enables effective continuous unlearning and successfully defends against relearning attacks.
  arXiv  Detail & Related papers  (2025-08-21T11:12:09Z)
• Value from Observations: Towards Large-Scale Imitation Learning via Self-Improvement [19.883973457999282]
  Imitation Learning from Observation (IfO) offers a powerful way to learn behaviors at large-scale.<n>This paper investigates idealized scenarios with mostly bimodal-quality data distributions and introduces a method to learn from such data.<n>Our method adapts RL-based imitation learning to action-free demonstrations, using a value function to transfer information between expert and non-expert data.
  arXiv  Detail & Related papers  (2025-07-09T09:55:23Z)
• Bridging Supervised Learning and Reinforcement Learning in Math Reasoning [55.889740979706815]
  Reinforcement Learning (RL) has played a central role in the recent surge of math abilities by enabling self-improvement through binary verifier signals.<n>In this work, we propose Negative-aware Fine-Tuning (NFT) -- a supervised approach that enables LLMs to reflect on their failures and improve autonomously with no external teachers.
  arXiv  Detail & Related papers  (2025-05-23T17:17:40Z)
• UniErase: Unlearning Token as a Universal Erasure Primitive for Language Models [54.75551043657238]
  We introduce UniErase, a novel unlearning paradigm that employs learnable parametric suffix (unlearning token) to steer language models toward targeted forgetting behaviors.<n>UniErase achieves state-of-the-art (SOTA) performance across batch, sequential, and precise unlearning under fictitious and real-world knowledge settings.
  arXiv  Detail & Related papers  (2025-05-21T15:53:28Z)
• Aligning Large Language Models to Follow Instructions and Hallucinate Less via Effective Data Filtering [66.5524727179286]
  Training LLMs on data that contains unfamiliar knowledge during the instruction tuning stage can make LLMs overconfident and encourage hallucinations.<n>We introduce a novel framework, NOVA, which identifies high-quality data that aligns well with the LLM's learned knowledge to reduce hallucinations.
  arXiv  Detail & Related papers  (2025-02-11T08:05:56Z)
• Clear Minds Think Alike: What Makes LLM Fine-tuning Robust? A Study of Token Perplexity [61.48338027901318]
  We show that fine-tuning with LLM-generated data improves target task performance and reduces out-of-domain degradation.<n>This is the first mechanistic explanation for the superior OOD robustness conferred by LLM-generated training data.
  arXiv  Detail & Related papers  (2025-01-24T08:18:56Z)
• Exploring Knowledge Boundaries in Large Language Models for Retrieval Judgment [56.87031484108484]
  Large Language Models (LLMs) are increasingly recognized for their practical applications. Retrieval-Augmented Generation (RAG) tackles this challenge and has shown a significant impact on LLMs. By minimizing retrieval requests that yield neutral or harmful results, we can effectively reduce both time and computational costs.
  arXiv  Detail & Related papers  (2024-11-09T15:12:28Z)
• Insights from the Inverse: Reconstructing LLM Training Goals Through Inverse RL [7.988692259455583]
  Large language models (LLMs) trained with Reinforcement Learning from Human Feedback have demonstrated remarkable capabilities, but their underlying reward functions and decision-making processes remain opaque. This paper introduces a novel approach to interpreting LLMs by applying inverse reinforcement learning (IRL) to recover their implicit reward functions. We conduct experiments on toxicity-aligned LLMs of varying sizes, extracting reward models that achieve up to 80.40% accuracy in predicting human preferences.
  arXiv  Detail & Related papers  (2024-10-16T12:14:25Z)
• Towards Robust Knowledge Unlearning: An Adversarial Framework for Assessing and Improving Unlearning Robustness in Large Language Models [19.015202590038996]
  We design Dynamic Unlearning Attack (DUA), a dynamic and automated framework to attack unlearned models. We propose Latent Adrial Unlearning (LAU), a universal framework that effectively enhances the robustness of the unlearned process. We demonstrate that LAU improves unlearning effectiveness by over $53.5%$, cause only less than a $11.6%$ reduction in neighboring knowledge, and have almost no impact on the model's general capabilities.
  arXiv  Detail & Related papers  (2024-08-20T09:36:04Z)
• Towards Effective Evaluations and Comparisons for LLM Unlearning Methods [97.2995389188179]
  This paper seeks to refine the evaluation of machine unlearning for large language models.<n>It addresses two key challenges -- the robustness of evaluation metrics and the trade-offs between competing goals.
  arXiv  Detail & Related papers  (2024-06-13T14:41:00Z)
• Uncertainty Aware Learning for Language Model Alignment [97.36361196793929]
  We propose uncertainty-aware learning (UAL) to improve the model alignment of different task scenarios. We implement UAL in a simple fashion -- adaptively setting the label smoothing value of training according to the uncertainty of individual samples. Experiments on widely used benchmarks demonstrate that our UAL significantly and consistently outperforms standard supervised fine-tuning.
  arXiv  Detail & Related papers  (2024-06-07T11:37:45Z)
• Large Language Models are Biased Reinforcement Learners [0.0]
  We show that large language models (LLMs) exhibit behavioral signatures of a relative value bias. Computational cognitive modeling reveals that LLM behavior is well-described by a simple RL algorithm.
  arXiv  Detail & Related papers  (2024-05-19T01:43:52Z)
• LLM In-Context Recall is Prompt Dependent [0.0]
  A model's ability to do this significantly influences its practical efficacy and dependability in real-world applications. This study demonstrates that an LLM's recall capability is not only contingent upon the prompt's content but also may be compromised by biases in its training data.
  arXiv  Detail & Related papers  (2024-04-13T01:13:59Z)
• Balancing Exploration and Exploitation in LLM using Soft RLLF for Enhanced Negation Understanding [4.799288023353623]
  Finetuning approaches in NLP often focus on exploitation rather than exploration, which may lead to suboptimal models. We leverage Reinforcement Learning from Logical Feedback to create an effective balance between exploration and exploitation in language models. This has implications for the development of more accurate, reliable, and logically consistent language models in high-stakes domains.
  arXiv  Detail & Related papers  (2024-03-02T11:54:55Z)
• Learning Objective-Specific Active Learning Strategies with Attentive Neural Processes [72.75421975804132]
  Learning Active Learning (LAL) suggests to learn the active learning strategy itself, allowing it to adapt to the given setting. We propose a novel LAL method for classification that exploits symmetry and independence properties of the active learning problem. Our approach is based on learning from a myopic oracle, which gives our model the ability to adapt to non-standard objectives.
  arXiv  Detail & Related papers  (2023-09-11T14:16:37Z)
• Improving Open Information Extraction with Large Language Models: A Study on Demonstration Uncertainty [52.72790059506241]
  Open Information Extraction (OIE) task aims at extracting structured facts from unstructured text. Despite the potential of large language models (LLMs) like ChatGPT as a general task solver, they lag behind state-of-the-art (supervised) methods in OIE tasks.
  arXiv  Detail & Related papers  (2023-09-07T01:35:24Z)
• Large Language Models Are Latent Variable Models: Explaining and Finding Good Demonstrations for In-Context Learning [104.58874584354787]
  In recent years, pre-trained large language models (LLMs) have demonstrated remarkable efficiency in achieving an inference-time few-shot learning capability known as in-context learning. This study aims to examine the in-context learning phenomenon through a Bayesian lens, viewing real-world LLMs as latent variable models.
  arXiv  Detail & Related papers  (2023-01-27T18:59:01Z)
• TRAIL: Near-Optimal Imitation Learning with Suboptimal Data [100.83688818427915]
  We present training objectives that use offline datasets to learn a factored transition model. Our theoretical analysis shows that the learned latent action space can boost the sample-efficiency of downstream imitation learning. To learn the latent action space in practice, we propose TRAIL (Transition-Reparametrized Actions for Imitation Learning), an algorithm that learns an energy-based transition model.
  arXiv  Detail & Related papers  (2021-10-27T21:05:00Z)
• DEALIO: Data-Efficient Adversarial Learning for Imitation from Observation [57.358212277226315]
  In imitation learning from observation IfO, a learning agent seeks to imitate a demonstrating agent using only observations of the demonstrated behavior without access to the control signals generated by the demonstrator. Recent methods based on adversarial imitation learning have led to state-of-the-art performance on IfO problems, but they typically suffer from high sample complexity due to a reliance on data-inefficient, model-free reinforcement learning algorithms. This issue makes them impractical to deploy in real-world settings, where gathering samples can incur high costs in terms of time, energy, and risk. We propose a more data-efficient IfO algorithm
  arXiv  Detail & Related papers  (2021-03-31T23:46:32Z)

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.