English
Related papers

Related papers: Safeguarding LLM Fine-tuning via Push-Pull Distrib…

200 papers

We introduce the Smoothed Online Optimization for Target Tracking (SOOTT) problem, a new framework that integrates three key objectives in online decision-making under uncertainty: (1) tracking cost for following a dynamically moving…

Machine Learning · Computer Science 2025-09-09 Ali Zeynali , Mahsa Sahebdel , Qingsong Liu , Mohammad Hajiesmaili , Ramesh K. Sitaraman

Post-training large language models (LLMs) often suffers from catastrophic forgetting, where improvements on a target objective degrade previously acquired capabilities. Recent evidence suggests that this phenomenon is primarily driven by…

Machine Learning · Computer Science 2026-05-07 Xinyu Wang , Changzhi Sun , Yuanbin Wu , Xiaoling Wang

Large language models (LLMs) have emerged as powerful tools but pose significant safety risks through harmful outputs and vulnerability to adversarial attacks. We propose SaP, short for Safety Polytope, a geometric approach to LLM safety…

Machine Learning · Computer Science 2025-06-02 Xin Chen , Yarden As , Andreas Krause

Large Language Models (LLMs) pose a significant risk of safety misalignment after finetuning, as models can be compromised by both explicitly and implicitly harmful data. Even some seemingly benign data can inadvertently steer a model…

Computation and Language · Computer Science 2026-05-15 Zhanhao Hu , Xiao Huang , Patrick Mendoza , Emad A. Alghamdi , Basel Alomair , Raluca Ada Popa , David Wagner

Handling distribution shifts from training data, known as out-of-distribution (OOD) generalization, poses a significant challenge in the field of machine learning. While a pre-trained vision-language model like CLIP has demonstrated…

Computer Vision and Pattern Recognition · Computer Science 2024-07-04 Bac Nguyen , Stefan Uhlich , Fabien Cardinaux , Lukas Mauch , Marzieh Edraki , Aaron Courville

The alignment problem in Large Language Models (LLMs) involves adapting them to the broad spectrum of human values. This requirement challenges existing alignment methods due to diversity of preferences and regulatory standards. This paper…

Computation and Language · Computer Science 2024-02-28 Xinyu Lu , Bowen Yu , Yaojie Lu , Hongyu Lin , Haiyang Yu , Le Sun , Xianpei Han , Yongbin Li

Alignment plays a fundamental role in many machine learning problems, such as multi-network analysis, multimodal learning, and point cloud registration. Recent works increasingly leverage optimal transport (OT) for distributional alignment,…

Machine Learning · Computer Science 2026-05-26 Qi Yu , Ruizhong Qiu , Zhichen Zeng , My T. Thai , Huan Liu , Hanghang Tong

Customizing Large Language Models (LLMs) on untrusted datasets poses severe risks of injecting toxic behaviors. In this work, we introduce Optimus, a novel defense framework designed to mitigate fine-tuning harms while preserving…

This paper studies the Partial Optimal Transport (POT) problem between two unbalanced measures with at most $n$ supports and its applications in various AI tasks such as color transfer or domain adaptation. There is hence the need for fast…

Machine Learning · Computer Science 2023-12-25 Anh Duc Nguyen , Tuan Dung Nguyen , Quang Minh Nguyen , Hoang H. Nguyen , Lam M. Nguyen , Kim-Chuan Toh

Recent advancements in large language models (LLMs) have underscored their vulnerability to safety alignment jailbreaks, particularly when subjected to downstream fine-tuning. However, existing mitigation strategies primarily focus on…

Cryptography and Security · Computer Science 2025-06-06 Lei Hsiung , Tianyu Pang , Yung-Chen Tang , Linyue Song , Tsung-Yi Ho , Pin-Yu Chen , Yaoqing Yang

Orthogonal parameter-efficient fine-tuning (PEFT) adapts pretrained weights through structure-preserving multiplicative transformations, but existing methods often conflate two distinct design choices: the subspace in which adaptation…

Machine Learning · Computer Science 2026-05-13 Lanxin Zhao , Bamdev Mishra , Pratik Jawanpuria , Lequan Lin , Dai Shi , Junbin Gao , Andi Han

Ensuring Large Language Model (LLM) safety is crucial, yet the lack of a clear understanding about safety mechanisms hinders the development of precise and reliable methodologies for safety intervention across diverse tasks. To better…

Cryptography and Security · Computer Science 2026-04-10 Weiwei Qi , Zefeng Wu , Tianhang Zheng , Zikang Zhang , Xiaojun Jia , Zhan Qin , Kui Ren

Supervised fine-tuning (SFT) is crucial in adapting large language model (LLMs) to a specific domain or task. However, only a limited amount of labeled data is available in practical applications, which poses a severe challenge for SFT in…

Computation and Language · Computer Science 2025-02-20 Junyu Luo , Xiao Luo , Xiusi Chen , Zhiping Xiao , Wei Ju , Ming Zhang

Large language models (LLMs) have become increasingly central to AI applications worldwide, necessitating robust multilingual safety alignment to ensure secure deployment across diverse linguistic contexts. Existing preference learning…

Computation and Language · Computer Science 2025-05-23 Weixiang Zhao , Yulin Hu , Yang Deng , Tongtong Wu , Wenxuan Zhang , Jiahe Guo , An Zhang , Yanyan Zhao , Bing Qin , Tat-Seng Chua , Ting Liu

Fine-tuning large pretrained language models (LLMs) is a cornerstone of modern NLP, yet its growing memory demands (driven by backpropagation and large optimizer States) limit deployment in resource-constrained settings. Zero-order (ZO)…

Machine Learning · Computer Science 2026-02-17 Valery Parfenov , Grigoriy Evseev , Andrey Veprikov , Nikolay Bushkov , Stanislav Moiseev , Aleksandr Beznosikov

Few-Shot classification aims at solving problems that only a few samples are available in the training process. Due to the lack of samples, researchers generally employ a set of training tasks from other domains to assist the target task,…

Computer Vision and Pattern Recognition · Computer Science 2022-04-04 Renjie Xu , Xinghao Yang , Baodi Liu , Kai Zhang , Weifeng Liu

Fine-tuning on task-specific data to boost downstream performance is a crucial step for leveraging Large Language Models (LLMs). However, previous studies have demonstrated that fine-tuning the models on several adversarial samples or even…

Machine Learning · Computer Science 2024-10-14 Han Shen , Pin-Yu Chen , Payel Das , Tianyi Chen

Despite the success of deep learning-based algorithms, it is widely known that neural networks may fail to be robust. A popular paradigm to enforce robustness is adversarial training (AT), however, this introduces many computational and…

Machine Learning · Computer Science 2024-01-18 Nicolas Garcia Trillos , Matt Jacobs , Jakwang Kim , Matthew Werenski

Reasoning-augmented Vision-Language Models (RVLMs) rely on safety alignment to prevent harmful behavior, yet their exposed chain-of-thought (CoT) traces introduce new attack surfaces. In this work, we find that the safety alignment of RVLMs…

Computation and Language · Computer Science 2026-03-10 Le Yu , Zhengyue Zhao , Yawen Zheng , Yunhao Liu

Safety alignment is critical for deploying large language models (LLMs) in real-world applications, yet most existing approaches rely on large human-annotated datasets and static red-teaming benchmarks that are costly, difficult to scale,…

Computation and Language · Computer Science 2026-03-10 Punyajoy Saha , Sudipta Halder , Debjyoti Mondal , Subhadarshi Panda