Machine Unlearning
Overview
Machine unlearning asks how to remove a person’s data, a class, or a learned concept from a model after training, often to support privacy, safety, or changing data rights. Approaches range from exact or certified deletion methods, such as data partitioning and noisy or influence-based updates, to approximate parameter edits, representation changes, and inference-time steering; some systems are designed from the outset to make deletion easier. Researchers evaluate methods against retrained models and test forgetting, retained utility, privacy attacks, recovery, and repeated deletion across language, vision, diffusion, graph, and multimodal systems. A central tension is that suppressed outputs do not necessarily mean information is erased, while stronger removal can damage related knowledge; scalable methods with reliable guarantees and low collateral harm remain a challenge.
Latest papers
372 papers34 papers in the last four weeks, down 40% on the four weeks before. 0.2% of all new papers.
Spectral Unlearning: Transformer Structure-Preserving Updates for Language Models
Unlearning at Speed: Rapid Activation Dampening for Efficient Machine Unlearning in Real-World Face Recognition
DISCERN: Distributional Synthetic Replay for Retain-Free Incremental Machine Unlearning
Do unlearning algorithms learn the right features?
IDSISA: Influence-Distilled Sharded Training for Efficient Machine Unlearning
Auditable Unlearning for LLMs
A Unified Framework for Training-Dataset Aggregation and Exact Machine Unlearning via Gibbs Probability Measures
Sample-Grained Approximate Unlearning with Provable Per-Sample Bounds
Error Accumulation in Second-Order Sequential Machine Unlearning
Adaptive Localization of Knowledge Negation for Continual LLM Unlearning
Unlearning by Extrapolating a Shrink-and-Heal Lookahead
What Should LLMs Forget: Reinforcement Unlearning of Knowledge Associations
Loss is Not Behavior: A Unified Output-Space Analysis of Gradient-Based Machine Unlearning
DecisionUnlearn-Bench: Unlearning Probabilistic Decisions under Changing Schemas
Forget the Data, Keep the Capability: Switchable Unlearning for Retrieval
Risk-Aware Output Control Should be the Focus of Machine Unlearning
Approximate Unlearning in Prototype-based-classifiers
Exact Unlearning After Data Curation
TRACE: Data-Free Text Reconstruction Attacks against Approximate Unlearning in LLMs
When Forgotten Facts Return: Relearning-Aware Unlearning of Knowledge Graph Embeddings via Tangent Kernels
CU in RAG: Compositional Concept Unlearning in Retrieval-Augmented Generation
Beyond the Frame: Redundancy, Not Trajectory, Carries the Privacy Leak in Video Unlearning
person) and (car) on VSPW and (car) on Virtual KITTI~2, achieving on the latter the strongest suppression among verdict-safe methods retaining usable utility, while retaining mIoU on VSPW car against at most for any baseline achieving comparable erasure.