WAGLE: Strategic Weight Attribution for Effective and Modular Unlearning in Large Language Models

The need for effective unlearning mechanisms in large language models (LLMs) is increasingly urgent, driven by the necessity to adhere to data regulations and foster ethical generative AI practices. Despite growing interest of LLM unlearning, much of the existing research has focused on varied unlearning method designs to boost effectiveness and efficiency. However, the inherent relationship between model weights and LLM unlearning has not been extensively examined. In this paper, we systematically explore how model weights interact with unlearning processes in LLMs and we design the weight attribution-guided LLM unlearning method, WAGLE, which unveils the interconnections between 'influence' of weights and 'influence' of data to forget and retain in LLM generation. By strategically guiding the LLM unlearning across different types of unlearning methods and tasks, WAGLE can erase the undesired content, while maintaining the performance of the original tasks. We refer to the weight attribution-guided LLM unlearning method as WAGLE, which unveils the interconnections between 'influence' of weights and 'influence' of data to forget and retain in LLM generation. Our extensive experiments show that WAGLE boosts unlearning performance across a range of LLM unlearning methods such as gradient difference and (negative) preference optimization, applications such as fictitious unlearning, malicious use prevention, and copyrighted information removal, and models including Zephyr-7b-beta and Llama2-7b. To the best of our knowledge, our work offers the first principled method for attributing and pinpointing the influential weights in enhancing LLM unlearning. It stands in contrast to previous methods that lack weight attribution and simpler weight attribution techniques.

LLM Self-Correction with DeCRIM: Decompose, Critique, and Refine for Enhanced Following of Instructions with Multiple Constraints

Instruction following is a key capability for LLMs. However, recent studies have shown that LLMs often struggle with instructions containing multiple constraints (e.g. a request to create a social media post "in a funny tone" with "no hashtag"). Despite this, most evaluations focus solely on synthetic data. To address this, we introduce RealInstruct, the first benchmark designed to evaluate LLMs' ability to follow real-world multi-constrained instructions by leveraging queries real users asked AI assistants. We also investigate model-based evaluation as a cost-effective alternative to human annotation for this task. Our findings reveal that even the proprietary GPT-4 model fails to meet at least one constraint on over 21% of instructions, highlighting the limitations of state-of-the-art models. To address the performance gap between open-source and proprietary models, we propose the Decompose, Critique and Refine (DeCRIM) self-correction pipeline, which enhances LLMs' ability to follow constraints. DeCRIM works by decomposing the original instruction into a list of constraints and using a Critic model to decide when and where the LLM's response needs refinement. Our results show that DeCRIM improves Mistral's performance by 7.3% on RealInstruct and 8.0% on IFEval even with weak feedback. Moreover, we demonstrate that with strong feedback, open-source LLMs with DeCRIM can outperform GPT-4 on both benchmarks.

Simplicity Prevails: Rethinking Negative Preference Optimization for LLM Unlearning

In this work, we address the problem of large language model (LLM) unlearning, aiming to remove unwanted data influences and associated model capabilities (e.g., copyrighted data or harmful content generation) while preserving essential model utilities, without the need for retraining from scratch. Despite the growing need for LLM unlearning, a principled optimization framework remains lacking. To this end, we revisit the state-of-the-art approach, negative preference optimization (NPO), and identify the issue of reference model bias, which could undermine NPO's effectiveness, particularly when unlearning forget data of varying difficulty. Given that, we propose a simple yet effective unlearning optimization framework, called SimNPO, showing that 'simplicity' in removing the reliance on a reference model (through the lens of simple preference optimization) benefits unlearning. We also provide deeper insights into SimNPO's advantages, supported by analysis using mixtures of Markov chains. Furthermore, we present extensive experiments validating SimNPO's superiority over existing unlearning baselines in benchmarks like TOFU and MUSE, and robustness against relearning attacks. Codes are available at https://github.com/OPTML-Group/Unlearn-Simple.

Pruning then Reweighting: Towards Data-Efficient Training of Diffusion Models

Despite the remarkable generation capabilities of Diffusion Models (DMs), conducting training and inference remains computationally expensive. Previous works have been devoted to accelerating diffusion sampling, but achieving data-efficient diffusion training has often been overlooked. In this work, we investigate efficient diffusion training from the perspective of dataset pruning. Inspired by the principles of data-efficient training for generative models such as generative adversarial networks (GANs), we first extend the data selection scheme used in GANs to DM training, where data features are encoded by a surrogate model, and a score criterion is then applied to select the coreset. To further improve the generation performance, we employ a class-wise reweighting approach, which derives class weights through distributionally robust optimization (DRO) over a pre-trained reference DM. For a pixel-wise DM (DDPM) on CIFAR-10, experiments demonstrate the superiority of our methodology over existing approaches and its effectiveness in image synthesis comparable to that of the original full-data model while achieving the speed-up between 2.34 times and 8.32 times. Additionally, our method could be generalized to latent DMs (LDMs), e.g., Masked Diffusion Transformer (MDT) and Stable Diffusion (SD), and achieves competitive generation capability on ImageNet.Code is available here (https://github.com/Yeez-lee/Data-Selection-and-Reweighting-for-Diffusion-Models).

Transient Adversarial 3D Projection Attacks on Object Detection in Autonomous Driving

Object detection is a crucial task in autonomous driving. While existing research has proposed various attacks on object detection, such as those using adversarial patches or stickers, the exploration of projection attacks on 3D surfaces remains largely unexplored. Compared to adversarial patches or stickers, which have fixed adversarial patterns, projection attacks allow for transient modifications to these patterns, enabling a more flexible attack. In this paper, we introduce an adversarial 3D projection attack specifically targeting object detection in autonomous driving scenarios. We frame the attack formulation as an optimization problem, utilizing a combination of color mapping and geometric transformation models. Our results demonstrate the effectiveness of the proposed attack in deceiving YOLOv3 and Mask R-CNN in physical settings. Evaluations conducted in an indoor environment show an attack success rate of up to 100% under low ambient light conditions, highlighting the potential damage of our attack in real-world driving scenarios.

Adversarial Watermarking for Face Recognition

Watermarking is an essential technique for embedding an identifier (i.e., watermark message) within digital images to assert ownership and monitor unauthorized alterations. In face recognition systems, watermarking plays a pivotal role in ensuring data integrity and security. However, an adversary could potentially interfere with the watermarking process, significantly impairing recognition performance. We explore the interaction between watermarking and adversarial attacks on face recognition models. Our findings reveal that while watermarking or input-level perturbation alone may have a negligible effect on recognition accuracy, the combined effect of watermarking and perturbation can result in an adversarial watermarking attack, significantly degrading recognition performance. Specifically, we introduce a novel threat model, the adversarial watermarking attack, which remains stealthy in the absence of watermarking, allowing images to be correctly recognized initially. However, once watermarking is applied, the attack is activated, causing recognition failures. Our study reveals a previously unrecognized vulnerability: adversarial perturbations can exploit the watermark message to evade face recognition systems. Evaluated on the CASIA-WebFace dataset, our proposed adversarial watermarking attack reduces face matching accuracy by 67.2% with an $\ell_\infty$ norm-measured perturbation strength of ${2}/{255}$ and by 95.9% with a strength of ${4}/{255}$.

Learning on Transformers is Provable Low-Rank and Sparse: A One-layer Analysis

Efficient training and inference algorithms, such as low-rank adaption and model pruning, have shown impressive performance for learning Transformer-based large foundation models. However, due to the technical challenges of the non-convex optimization caused by the complicated architecture of Transformers, the theoretical study of why these methods can be applied to learn Transformers is mostly elusive. To the best of our knowledge, this paper shows the first theoretical analysis of the property of low-rank and sparsity of one-layer Transformers by characterizing the trained model after convergence using stochastic gradient descent. By focusing on a data model based on label-relevant and label-irrelevant patterns, we quantify that the gradient updates of trainable parameters are low-rank, which depends on the number of label-relevant patterns. We also analyze how model pruning affects the generalization while improving computation efficiency and conclude that proper magnitude-based pruning has a slight effect on the testing performance. We implement numerical experiments to support our findings.

Reversing the Forget-Retain Objectives: An Efficient LLM Unlearning Framework from Logit Difference

As Large Language Models (LLMs) demonstrate extensive capability in learning from documents, LLM unlearning becomes an increasingly important research area to address concerns of LLMs in terms of privacy, copyright, etc. A conventional LLM unlearning task typically involves two goals: (1) The target LLM should forget the knowledge in the specified forget documents, and (2) it should retain the other knowledge that the LLM possesses, for which we assume access to a small number of retain documents. To achieve both goals, a mainstream class of LLM unlearning methods introduces an optimization framework with a combination of two objectives - maximizing the prediction loss on the forget documents while minimizing that on the retain documents, which suffers from two challenges, degenerated output and catastrophic forgetting. In this paper, we propose a novel unlearning framework called Unlearning from Logit Difference (ULD), which introduces an assistant LLM that aims to achieve the opposite of the unlearning goals: remembering the forget documents and forgetting the retain knowledge. ULD then derives the unlearned LLM by computing the logit difference between the target and the assistant LLMs. We show that such reversed objectives would naturally resolve both aforementioned challenges while significantly improving the training efficiency. Extensive experiments demonstrate that our method efficiently achieves the intended forgetting while preserving the LLM's overall capabilities, reducing training time by more than threefold. Notably, our method loses 0% of model utility on the ToFU benchmark, whereas baseline methods may sacrifice 17% of utility on average to achieve comparable forget quality. Our code will be publicly available at https://github.com/UCSB-NLP-Chang/ULD.

Label Smoothing Improves Machine Unlearning

The objective of machine unlearning (MU) is to eliminate previously learned data from a model. However, it is challenging to strike a balance between computation cost and performance when using existing MU techniques. Taking inspiration from the influence of label smoothing on model confidence and differential privacy, we propose a simple gradient-based MU approach that uses an inverse process of label smoothing. This work introduces UGradSL, a simple, plug-and-play MU approach that uses smoothed labels. We provide theoretical analyses demonstrating why properly introducing label smoothing improves MU performance. We conducted extensive experiments on six datasets of various sizes and different modalities, demonstrating the effectiveness and robustness of our proposed method. The consistent improvement in MU performance is only at a marginal cost of additional computations. For instance, UGradSL improves over the gradient ascent MU baseline by 66% unlearning accuracy without sacrificing unlearning efficiency.

PSBD: Prediction Shift Uncertainty Unlocks Backdoor Detection

Deep neural networks are susceptible to backdoor attacks, where adversaries manipulate model predictions by inserting malicious samples into the training data. Currently, there is still a lack of direct filtering methods for identifying suspicious training data to unveil potential backdoor samples. In this paper, we propose a novel method, Prediction Shift Backdoor Detection (PSBD), leveraging an uncertainty-based approach requiring minimal unlabeled clean validation data. PSBD is motivated by an intriguing Prediction Shift (PS) phenomenon, where poisoned models' predictions on clean data often shift away from true labels towards certain other labels with dropout applied during inference, while backdoor samples exhibit less PS. We hypothesize PS results from neuron bias effect, making neurons favor features of certain classes. PSBD identifies backdoor training samples by computing the Prediction Shift Uncertainty (PSU), the variance in probability values when dropout layers are toggled on and off during model inference. Extensive experiments have been conducted to verify the effectiveness and efficiency of PSBD, which achieves state-of-the-art results among mainstream detection methods.