English
Related papers

Related papers: MBTSAD: Mitigating Backdoors in Language Models Ba…

200 papers

The advent of Large Language Models (LLMs) has marked significant achievements in language processing and reasoning capabilities. Despite their advancements, LLMs face vulnerabilities to data poisoning attacks, where the adversary inserts…

Machine Learning · Computer Science 2025-05-30 Xiangyu Zhou , Yao Qiang , Saleh Zare Zade , Mohammad Amin Roshani , Prashant Khanduri , Douglas Zytko , Dongxiao Zhu

Deep neural networks have played a crucial part in many critical domains, such as autonomous driving, face recognition, and medical diagnosis. However, deep neural networks are facing security threats from backdoor attacks and can be…

Cryptography and Security · Computer Science 2023-11-30 Jiyang Guan , Jian Liang , Ran He

Token dropping is a recently-proposed strategy to speed up the pretraining of masked language models, such as BERT, by skipping the computation of a subset of the input tokens at several middle layers. It can effectively reduce the training…

Computation and Language · Computer Science 2023-05-25 Qihuang Zhong , Liang Ding , Juhua Liu , Xuebo Liu , Min Zhang , Bo Du , Dacheng Tao

The use of third-party datasets and pre-trained machine learning models poses a threat to NLP systems due to possibility of hidden backdoor attacks. Existing attacks involve poisoning the data samples such as insertion of tokens or sentence…

Computation and Language · Computer Science 2024-04-09 Irina Alekseevskaia , Konstantin Arkhipenko

Fine-Tuning-as-a-Service (FTaaS) facilitates the customization of Multimodal Large Language Models (MLLMs) but introduces critical backdoor risks via poisoned data. Existing defenses either rely on supervised signals or fail to generalize…

Artificial Intelligence · Computer Science 2026-05-25 Mingzu Liu , Hao Fang , Runmin Cong

Because state-of-the-art language models are expensive to train, most practitioners must make use of one of the few publicly available language models or language model APIs. This consolidation of trust increases the potency of backdoor…

Cryptography and Security · Computer Science 2023-07-28 Nikhil Kandpal , Matthew Jagielski , Florian Tramèr , Nicholas Carlini

Over the past few years, the emergence of backdoor attacks has presented significant challenges to deep learning systems, allowing attackers to insert backdoors into neural networks. When data with a trigger is processed by a backdoor…

Cryptography and Security · Computer Science 2025-03-07 Haiyang Yu , Tian Xie , Jiaping Gui , Pengyang Wang , Ping Yi , Yue Wu

Language models significantly benefit from context tokens, such as prompts or scratchpads. They perform better when prompted with informative instructions, and they acquire new reasoning capabilities by generating a scratch-pad before…

Computation and Language · Computer Science 2022-10-03 Charlie Snell , Dan Klein , Ruiqi Zhong

Recent studies have revealed that \textit{Backdoor Attacks} can threaten the safety of natural language processing (NLP) models. Investigating the strategies of backdoor attacks will help to understand the model's vulnerability. Most…

Machine Learning · Computer Science 2023-10-26 Weimin Lyu , Songzhu Zheng , Lu Pang , Haibin Ling , Chao Chen

Backdoor attacks impose a new threat in Deep Neural Networks (DNNs), where a backdoor is inserted into the neural network by poisoning the training dataset, misclassifying inputs that contain the adversary trigger. The major challenge for…

Machine Learning · Computer Science 2024-09-26 Yue Wang , Wenqing Li , Esha Sarkar , Muhammad Shafique , Michail Maniatakos , Saif Eddin Jabari

Multi-modal large language models (MLLMs) extend large language models (LLMs) to process multi-modal information, enabling them to generate responses to image-text inputs. MLLMs have been incorporated into diverse multi-modal applications,…

Cryptography and Security · Computer Science 2025-03-21 Zenghui Yuan , Jiawen Shi , Pan Zhou , Neil Zhenqiang Gong , Lichao Sun

Backdoor defenses have been studied to alleviate the threat of deep neural networks (DNNs) being backdoor attacked and thus maliciously altered. Since DNNs usually adopt some external training data from an untrusted third party, a robust…

Computer Vision and Pattern Recognition · Computer Science 2023-03-24 Kuofeng Gao , Yang Bai , Jindong Gu , Yong Yang , Shu-Tao Xia

Multimodal Diffusion Language Models (MDLMs) have recently emerged as a competitive alternative to their autoregressive counterparts. Yet their vulnerability to backdoor attacks remains largely unexplored. In this work, we show that…

Cryptography and Security · Computer Science 2026-02-27 Guangnian Wan , Qi Li , Gongfan Fang , Xinyin Ma , Xinchao Wang

Backdoor attacks represent a subtle yet effective class of cyberattacks targeting AI models, primarily due to their stealthy nature. The model behaves normally on clean data but exhibits malicious behavior only when the attacker embeds a…

Machine Learning · Computer Science 2025-09-29 Sujeevan Aseervatham , Achraf Kerzazi , Younès Bennani

Supervised fine-tuning (SFT) is widely used to inject new knowledge into language models, but it often degrades pretrained capabilities such as reasoning and general-domain performance. We argue this forgetting arises because fine-tuning…

Computation and Language · Computer Science 2026-05-22 Jiarui Liu , Lechen Zhang , Yongjin Yang , Yinghui He , Yingheng Wang , Weihao Xuan , Zhijing Jin , Mona Diab

Masked diffusion language models (MDLMs) are emerging as a compelling new paradigm for text generation, but their training-time security remains largely unexplored. Existing backdoor attacks on Gaussian diffusion models or autoregressive…

Machine Learning · Computer Science 2026-05-20 Daniel Yiming Cao , Chengzhong Wang , Sheng-Yen Chou , Chengyu Huang , Pin-Yu Chen , Shengwei An

The rapid growth of natural language processing (NLP) and pre-trained language models have enabled accurate text classification in a variety of settings. However, text classification models are susceptible to backdoor attacks, where an…

Cryptography and Security · Computer Science 2024-12-30 A. Dilara Yavuz , M. Emre Gursoy

In-context learning, a paradigm bridging the gap between pre-training and fine-tuning, has demonstrated high efficacy in several NLP tasks, especially in few-shot settings. Despite being widely applied, in-context learning is vulnerable to…

Computation and Language · Computer Science 2024-10-10 Shuai Zhao , Meihuizi Jia , Luu Anh Tuan , Fengjun Pan , Jinming Wen

Despite the notable success of language models (LMs) in various natural language processing (NLP) tasks, the reliability of LMs is susceptible to backdoor attacks. Prior research attempts to mitigate backdoor learning while training the LMs…

Computation and Language · Computer Science 2024-06-04 Zongru Wu , Zhuosheng Zhang , Pengzhou Cheng , Gongshen Liu

The security threat of backdoor attacks is a central concern for deep neural networks (DNNs). Recently, without poisoned data, unlearning models with clean data and then learning a pruning mask have contributed to backdoor defense.…

Cryptography and Security · Computer Science 2024-05-31 Weilin Lin , Li Liu , Shaokui Wei , Jianze Li , Hui Xiong