Shang Liu

Zhenyu Yin, Shang Liu, Guangyuan Xu

The increasing number of Distributed Denial of Service (DDoS) attacks poses a major threat to the Internet, highlighting the importance of DDoS mitigation. Most existing approaches require complex training methods to learn data features, which increases the complexity and generality of the application. In this paper, we propose DrLLM, which aims to mine anomalous traffic information in zero-shot scenarios through Large Language Models (LLMs). To bridge the gap between DrLLM and existing approaches, we embed the global and local information of the traffic data into the reasoning paradigm and design three modules, namely Knowledge Embedding, Token Embedding, and Progressive Role Reasoning, for data representation and reasoning. In addition we explore the generalization of prompt engineering in the cybersecurity domain to improve the classification capability of DrLLM. Our ablation experiments demonstrate the applicability of DrLLM in zero-shot scenarios and further demonstrate the potential of LLMs in the network domains. DrLLM implementation code has been open-sourced at https://github.com/liuup/DrLLM.

Hao Du, Shang Liu, Lele Zheng et al.

Fine-tuning has emerged as a critical process in leveraging Large Language Models (LLMs) for specific downstream tasks, enabling these models to achieve state-of-the-art performance across various domains. However, the fine-tuning process often involves sensitive datasets, introducing privacy risks that exploit the unique characteristics of this stage. In this paper, we provide a comprehensive survey of privacy challenges associated with fine-tuning LLMs, highlighting vulnerabilities to various privacy attacks, including membership inference, data extraction, and backdoor attacks. We further review defense mechanisms designed to mitigate privacy risks in the fine-tuning phase, such as differential privacy, federated learning, and knowledge unlearning, discussing their effectiveness and limitations in addressing privacy risks and maintaining model utility. By identifying key gaps in existing research, we highlight challenges and propose directions to advance the development of privacy-preserving methods for fine-tuning LLMs, promoting their responsible use in diverse applications.

Shang Liu, Yu Pan, Guanting Chen et al.

Learning a reward model (RM) from human preferences has been an important component in aligning large language models (LLMs). The canonical setup of learning RMs from pairwise preference data is rooted in the classic Bradley-Terry (BT) model that accepts binary feedback, i.e., the label being either Response 1 is better than Response 2, or the opposite. Such a setup inevitably discards potentially useful samples (such as "tied" between the two responses) and loses more fine-grained information (such as "slightly better"). In this paper, we propose a framework for learning RMs under ordinal feedback which generalizes the case of binary preference feedback to any arbitrary granularity. Specifically, we first identify a marginal unbiasedness condition, which generalizes the assumption of the BT model in the existing binary feedback setting. The condition validates itself via the sociological concept of the wisdom of the crowd. Under the condition, we develop a natural probability model for pairwise preference data under ordinal feedback and analyze its properties. We prove the statistical benefits of ordinal feedback in terms of reducing the Rademacher complexity compared to the case of binary feedback. The proposed learning objective and the theory also extend to hinge loss and direct policy optimization (DPO). In particular, the theoretical analysis may be of independent interest when applying to a seemingly unrelated problem of knowledge distillation to interpret the bias-variance trade-off therein. The framework also sheds light on writing guidance for human annotators. Our numerical experiments validate that fine-grained feedback leads to better reward learning for both in-distribution and out-of-distribution settings. Further experiments show that incorporating a certain proportion of samples with tied preference boosts RM learning.