Understanding and shaping the behaviour of Large Language Models (LLMs) is increasingly important as applications become more powerful and more frequently adopted. This paper introduces a machine unlearning method specifically designed for LLMs. We introduce a selective pruning method for LLMs that removes neurons based on their relative importance on a targeted capability compared to overall network performance. This approach is a compute- and data-efficient method for identifying and removing neurons that enable specific behaviours. Our findings reveal that both feed-forward and attention neurons in LLMs are specialized; that is, for specific tasks, certain neurons are more crucial than others. Code from all experiments is available at https://github.com/nickypro/selective-pruning
Dissecting Fine-Tuning Unlearning in Large Language Models
International Digital Economy Academy (IDEA)（国际数字经济学院（IDEA））；South China University of Technology（华南理工大学）；Columbia University（哥伦比亚大学）；King Abdullah University of Science and Technology（阿卜杜拉国王科技大学）
Hong Kong Polytechnic University（香港理工大学）；Dalian University of Technology（大连理工大学）
We explore the problem of selectively forgetting categories from trained CNN classification models in the federated learning (FL). Given that the data used for training cannot be accessed globally in FL, our insights probe deep into the internal influence of each channel. Through the visualization of feature maps activated by different channels, we observe that different channels have a varying contribution to different categories in image classification. Inspired by this, we propose a method for scrubbing the model clean of information about particular categories. The method does not require …
U-CAN: Utility-Aware Contrastive Attenuation for Efficient Unlearning in Generative Recommendation
Guilin University of Electronic Technology（桂林电子科技大学）；University of Southampton（南安普顿大学）；Jinan University（暨南大学）；Nanjing University of Science and Technology（南京理工大学）
Indian Institute of Information Technology, Bhopal, India（印度比哈尔理工学院）；University of California, Santa Cruz, USA（加州大学圣克鲁兹分校）；Independent Researcher（独立研究者）；Stanford University, USA（斯坦福大学）；BITS Pilani Goa, India（比斯拉米印度学院）
Michigan State University（密歇根州立大学）；IBM Research（IBM研究院）；Cisco Research（思科研发）；University of California, Santa Cruz（加州大学圣克鲁兹分校）；ByteDance Research（字节跳动研究）；Carnegie Mellon University（卡内基梅隆大学）
Towards Efficient Target-Level Machine Unlearning Based on Essential Graph
University of Technology Sydney（悉尼科技大学）；City University of Macau（澳门城市大学）；Shenzhen Institute of Advanced Technology, University of Chinese Academy of Sciences（中国科学院大学深圳先进技术研究院）
State Key Laboratory of Integrated Services Networks (ISN)（综合业务网理论及关键技术国家重点实验室）；Shaanxi Key Laboratory of Network and System Security, Xidian University（西安电子科技大学陕西省网络与系统安全重点实验室）；School of Data Science and Engineering, East China Normal University（华东师范大学数据科学与工程学院）；Ant Group（蚂蚁集团）；College of Computer and Data Science, Fuzhou University（福州大学计算机与数据科学学院）
Backdoor injection attack is an emerging threat to the security of neural networks, however, there still exist limited effective defense methods against the attack. In this paper, we propose BAERASE, a novel method that can erase the backdoor injected into the victim model through machine unlearning. Specifically, BAERASE mainly implements backdoor defense in two key steps. First, trigger pattern recovery is conducted to extract the trigger patterns infected by the victim model. Here, the trigger pattern recovery problem is equivalent to the one of extracting an unknown noise distribution from…
Recent work on the lottery ticket hypothesis has produced highly sparse Transformers for NMT while maintaining BLEU. However, it is unclear how such pruning techniques affect a model's learned representations. By probing Transformers with more and more low-magnitude weights pruned away, we find that complex semantic information is first to be degraded. Analysis of internal activations reveals that higher layers diverge most over the course of pruning, gradually becoming less complex than their dense counterparts. Meanwhile, early layers of sparse models begin to perform more encoding. Attentio…
Roots Beneath the Cut: Uncovering the Risk of Concept Revival in Pruning-Based Unlearning for Diffusion Models
University of Georgia（佐治亚大学）；Carnegie Mellon University（卡内基梅隆大学）；Northeastern University（东北大学）；Stevens Institute of Technology（史蒂文斯理工学院）；University of Arizona（亚利桑那大学）
Pruning is a standard technique for removing unnecessary structure from a neural network to reduce its storage footprint, computational demands, or energy consumption. Pruning can reduce the parameter-counts of many state-of-the-art neural networks by an order of magnitude without compromising accuracy, meaning these networks contain a vast amount of unnecessary structure. In this paper, we study the relationship between pruning and interpretability. Namely, we consider the effect of removing unnecessary structure on the number of hidden units that learn disentangled representations of human-r…
Reconstructive Neuron Pruning for Backdoor Defense
Xidian University（西安电子科技大学）；Fudan University（复旦大学）；Shanghai Artificial Intelligence Laboratory（上海人工智能实验室）；University of Copenhagen（哥本哈根大学）；Sony AI（索尼人工智能）；University of Illinois at Urbana–Champaign（伊利诺伊大学厄巴纳-香槟分校）
