Saved in:
Bibliographic Details
Main Authors: Garg, Arpit, Saratchandran, Hemanth, Garg, Ravi, Lucey, Simon
Format: Preprint
Published: 2025
Subjects:
Online Access:https://arxiv.org/abs/2509.24166
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866915870389829632
author Garg, Arpit
Saratchandran, Hemanth
Garg, Ravi
Lucey, Simon
author_facet Garg, Arpit
Saratchandran, Hemanth
Garg, Ravi
Lucey, Simon
contents Machine unlearning in foundation models (e.g., language and vision transformers) is essential for privacy and safety; however, existing approaches are unstable and unreliable. A widely used strategy, the gradient difference method, applies gradient descent to retained data while performing gradient ascent on forgotten data. When combined with cross-entropy, this procedure can trigger the unbounded growth of weights and gradients, degrading both forgetting and retention. We provide a theoretical framework that explains this failure by showing how ascent destabilizes optimization in transformer feedforward MLP layers. Guided by this insight, we propose *Bounded Parameter-Efficient Unlearning*, which stabilizes LoRA-based fine-tuning by applying bounded functions to MLP adapters. This controls the weight dynamics during ascent and enables reliable convergence. We validate the approach on Vision Transformer class deletion on CIFAR-100, where GD+Sine is the only evaluated method to achieve both high forget quality and model utility across ViT-B/16, ViT-L/14, and DeiT-S architectures, and demonstrate generality on language-model benchmarks (TOFU, TDEC, MUSE) across architectures from 22M to 8B parameters, achieving improved forgetting while preserving utility.
format Preprint
id arxiv_https___arxiv_org_abs_2509_24166
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Stable Forgetting: Bounded Parameter-Efficient Unlearning in Foundation Models
Garg, Arpit
Saratchandran, Hemanth
Garg, Ravi
Lucey, Simon
Machine Learning
Artificial Intelligence
Machine unlearning in foundation models (e.g., language and vision transformers) is essential for privacy and safety; however, existing approaches are unstable and unreliable. A widely used strategy, the gradient difference method, applies gradient descent to retained data while performing gradient ascent on forgotten data. When combined with cross-entropy, this procedure can trigger the unbounded growth of weights and gradients, degrading both forgetting and retention. We provide a theoretical framework that explains this failure by showing how ascent destabilizes optimization in transformer feedforward MLP layers. Guided by this insight, we propose *Bounded Parameter-Efficient Unlearning*, which stabilizes LoRA-based fine-tuning by applying bounded functions to MLP adapters. This controls the weight dynamics during ascent and enables reliable convergence. We validate the approach on Vision Transformer class deletion on CIFAR-100, where GD+Sine is the only evaluated method to achieve both high forget quality and model utility across ViT-B/16, ViT-L/14, and DeiT-S architectures, and demonstrate generality on language-model benchmarks (TOFU, TDEC, MUSE) across architectures from 22M to 8B parameters, achieving improved forgetting while preserving utility.
title Stable Forgetting: Bounded Parameter-Efficient Unlearning in Foundation Models
topic Machine Learning
Artificial Intelligence
url https://arxiv.org/abs/2509.24166