PECKER: A Precisely Efficient Critical Knowledge Erasure Recipe For Machine Unlearning in Diffusion Models

Zhiyong Ma; Zhitao Deng; Huan Tang; Jialin Chen; Zhijun Zheng; Zhengping Li; Qingyuan Chuai

cs.AIApr 7, 2026

PECKER: A Precisely Efficient Critical Knowledge Erasure Recipe For Machine Unlearning in Diffusion Models

Authors: Zhiyong Ma, Zhitao Deng, Huan Tang, Jialin Chen, Zhijun Zheng, Zhengping Li, Qingyuan Chuai

Organizations: Cao Tu Li(Guangzhou) Technology Co., Ltd, China · South China University of Technology, China · Guangzhou Xinhua University, China · Hong Kong Baptist University, HongKong

Abstract

Machine unlearning (MU) has become a critical technique for GenAI models' safe and compliant operation. While existing MU methods are effective, most impose prohibitive training time and computational overhead. Our analysis suggests the root cause lies in poorly directed gradient updates, which reduce training efficiency and destabilize convergence. To mitigate these issues, we propose PECKER, an efficient MU approach that matches or outperforms prevailing methods. Within a distillation framework, PECKER introduces a saliency mask to prioritize updates to parameters that contribute most to forgetting the targeted data, thereby reducing unnecessary gradient computation and shortening overall training time without sacrificing unlearning efficacy. Our method generates samples that unlearn related class or concept more quickly, while closely aligning with the true image distribution on CIFAR-10 and STL-10 datasets, achieving shorter training times for both class forgetting and concept forgetting.

Figures & tables

Figure 1: Overview of PECKER. (1) A pseudo-score network $s_{\psi}$ provides data-free supervision; gradients are stopped through $s_{\psi}$ during generator updates. (2) We compute saliency $g_{s}$ and a mask $m_{s}$ from a temporarily frozen generator (Eq. 9 – 10 ), where red marks masked parameters. (3) Retain batches use full updates, while forget batches update only masked parameters. (4) The resulting generator produces concept-erased outputs for the forget condition.

Dataset	Method	UA( $\uparrow$ )	FID( $\downarrow$ )	IS( $\uparrow$ )	Precision( $\uparrow$ )	Data-free
CIFAR-10	Retrain	99.94	303.34	1.58	0.3834	$\times$
	SA [ 10 ]	100.00	13.80	8.19	0.3350	$\surd$
	SalUn [ 6 ]	-	14.99	6.09	0.3834	$\times$
	SFD [ 4 ]	93.72	5.71	9.03	0.6443	$\surd$
	PECKER	98.76	5.81	8.97	0.6454	$\surd$
STL-10	Retrain	97.32	273.93	1.73	0.0085	$\times$

Table 1: Experimental results on CIFAR-10 and STL-10 datasets. The bolded values denote the best score in each column, while the underlined indicate the second-best.

Panel of Figure 2 — Figure 2: Sample images during PECKER training. Top: forgetting trajectory for class 0 (plane), gradually replaced by class 1 (automobile in CIFAR-10 or bird in STL-10) across checkpoints. Bottom: $2\times 5$ grids of samples from retained classes at different training steps, where class labels are ordered from 1 to 9 (left-to-right, top-to-bottom) and random seeds are fixed across all grids for consistency.

Model	Brad Pitt		Angelina Jolie
Model	Prop. w/o Faces ( $\downarrow$ )	GCD ( $\downarrow$ )	Prop. w/o Faces ( $\downarrow$ )	GCD ( $\downarrow$ )
SD v1.4 [ 30 ]	10.4%	60.6%	11.7%	73.8%
SLD Medium [ 34 ]	14.1%	0.47%	11.9%	3.29%
ESD-x [ 8 ]	34.7%	2.01%	32.6%	3.35%
SA [ 10 ]	5.8%	7.52%	4.4%	7.74%
SFD-Two Stage [ 4 ]	1.76%	2.5%	1.92%	1.06%

Table 2: Quantitative results of celebrity forgetting tasks. The bolded values denote the best score in each column, while the underlined indicate the second-best.

Panel of Figure 3 — Figure 3: The curves of UA and FID for PECKER and SFD in 50,000 training steps on CIFAR-10. The blue lines represent PECKER, while the orange lines correspond to SFD.

Panel of Figure 4 — Figure 4: Celebrity forgetting process. Each column uses the same prompt and random seed at different checkpoints (0/20/40/60/80/100k imgs).

Figure 5: Generated images using various text-to-image diffusion models with prompts formulated as ‘A photo of a <nudity keyword> <human subject>.‘ Sensitive parts are manually censored after generation.

Figure 6: The process of Nudity Forgetting between PECKER and SFD. Due to the difficulty in displaying the entire nudity forgetting process over 300k images, and to balance image quality with the demonstration of the forgetting effects of our method, we selected 0, 20, 60, 80, 100k images to visually compare the entire process. Sensitive parts are manually censored after generation.

Machine Unlearning

Machine unlearning removes the influence of selected data or concepts from trained models while preserving useful behavior and avoiding full retraining.

See all 351 papers

Compensation-free Machine Unlearning in Text-to-Image Diffusion Models by Eliminating the Mutual Information
Mar 1, 2026Xinwen Cheng, Jingyuan Zhang, Zhehao Huang +2Diffusion Model Unlearning Concept Erasure
Institute of Image Processing and Pattern Recognition, School of Automation and Intelligent Sensing, Shanghai Jiao Tong University
Concept Acupuncturist: An Efficient and Precise Concept Erasure Method for Diffusion Models via Sparse Autoencoders
Oct 23, 2026Yiran Xiahou, Siliang Qin, Kai Chen +2Diffusion Model Unlearning Concept Erasure
Institute of Information Engineering, Chinese Academy of Sciences, China · Department of Computer Science, Metropolitan College, Boston University, USA · Institute of Data Science, NationalUniversity of SingaporeSingapore, Singapore
PURGE: Projected Unlearning via Retain-Guided Erasure
Jun 2, 2026Vedant Jawandhia, Daksh Ahuja, Ghufran Alam Siddiqui +3Privacy-Preserving ML Gradient Projection
1BITS Pilani, Pilani Campus, India

Mar 1, 2026J/K move · Enter open · S save

Compensation-free Machine Unlearning in Text-to-Image Diffusion Models by Eliminating the Mutual Information

Xinwen Cheng, Jingyuan Zhang, Zhehao Huang, Yingwen Wu+1

Institute of Image Processing and Pattern Recognition, School of Automation and Intelligent Sensing, Shanghai Jiao Tong University

Diffusion Model Unlearning Concept Erasure T2I Diffusion Models

Open paper AskPDF arXiv

The powerful generative capabilities of diffusion models have raised growing privacy and safety concerns regarding generating sensitive or undesired content. In response, machine unlearning (MU) -- commonly referred to as concept erasure (CE) in diffusion models -- has been introduced to remove specific knowledge from model parameters meanwhile preserving innocent knowledge. Despite recent advancements, existing unlearning methods often suffer from excessive and indiscriminate removal, which leads to substantial degradation in the quality of innocent generations. To preserve model utility, prior works rely on compensation, i.e., re-assimilating a subset of the remaining data or explicitly constraining the divergence from the pre-trained model on remaining concepts. However, we reveal that generations beyond the compensation scope still suffer, suggesting such post-remedial compensations are inherently insufficient for preserving the general utility of large-scale generative models. Therefore, in this paper, we advocate for developing compensation-free concept erasure operations, which precisely identify and eliminate the undesired knowledge such that the impact on other generations is minimal. In technique, we propose to MiM-MU, which is to unlearn a concept by minimizing the mutual information with a delicate design for computational effectiveness and for maintaining sampling distribution for other concepts. Extensive evaluations demonstrate that our proposed method achieves effective concept removal meanwhile maintaining high-quality generations for other concepts, and remarkably, without relying on any post-remedial compensation for the first time.

Abstract

Machine Unlearning

Explore similar work

Compensation-free Machine Unlearning in Text-to-Image Diffusion Models by Eliminating the Mutual Information

Concept Acupuncturist: An Efficient and Precise Concept Erasure Method for Diffusion Models via Sparse Autoencoders

PURGE: Projected Unlearning via Retain-Guided Erasure