# Feature-Selective Representation Misdirection for Machine Unlearning

Source: [Feature-Selective Representation Misdirection for Machine Unlearning](<https://www.sota2.com/research/paper/6944dbef423ea422657299b3>)

Collection index: [llms.txt](<https://www.sota2.com/research/llms.txt>)

Data updated: 2025-12-20T05:00:04.301Z

Authors: Taozhao Chen, Linghan Huang, Kim-Kwang Raymond Choo, Huaming Chen

Published: 2025-12-18T08:31:50.000Z

## Abstract

As large language models (LLMs) are increasingly adopted in safety-critical and regulated sectors, the retention of sensitive or prohibited knowledge introduces escalating risks, ranging from privacy leakage to regulatory non-compliance to to potential misuse, and so on. Recent studies suggest that machine unlearning can help ensure deployed models comply with evolving legal, safety, and governance requirements. However, current unlearning techniques assume clean separation between forget and retain datasets, which is challenging in operational settings characterized by highly entangled distributions. In such scenarios, perturbation-based methods often degrade general model utility or fail to ensure safety. To address this, we propose Selective Representation Misdirection for Unlearning (SRMU), a novel principled activation-editing framework that enforces feature-aware and directionally controlled perturbations. Unlike indiscriminate model weights perturbations, SRMU employs a structured misdirection vector with an activation importance map. The goal is to allow SRMU selectively suppresses harmful representations while preserving the utility on benign ones. Experiments are conducted on the widely used WMDP benchmark across low- and high-entanglement configurations. Empirical results reveal that SRMU delivers state-of-the-art unlearning performance with minimal utility losses, and remains effective under 20-30\\% overlap where existing baselines collapse. SRMU provides a robust foundation for safety-driven model governance, privacy compliance, and controlled knowledge removal in the emerging LLM-based applications. We release the replication package at https://figshare.com/s/d5931192a8824de26aff.

## Recorded benchmark results

A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage.

- [Machine Unlearning on WMDP (Bio, Cyber, Avg)](<https://www.sota2.com/research/sota/machine-unlearning-on-wmdp-bio-cyber-avg.md>): Machine Unlearning; WMDP; Bio Score: 28.5.

- [Language Understanding on MMLU (Accuracy)](<https://www.sota2.com/research/sota/language-understanding-on-mmlu-accuracy.md>): Language Understanding; MMLU; Accuracy: 57.1.

## Original sources and implementation

[Original paper](<https://arxiv.org/abs/2512.16297>)

[PDF](<https://arxiv.org/pdf/2512.16297.pdf>)
