arXiv subjects

Haihui Fan

Publications and source records attributed to Haihui Fan.

5 recordsLinked to original sources

arXiv2026

CrypFormBench: Benchmarking Formal Analysis Capability of Large Language Models for Cryptographic Schemes

Manual formal analysis of cryptographic schemes is labor-intensive and requires substantial expertise. While model-checking tools (e.g., Scyther and Tamarin) and computational-security tools (e.g., CryptoVerif and EasyCrypt) improve the automation of security proofs, they still rely on experts to abstract schemes and write tool-specific formal descriptions. Large language models (LLMs) are a promising alternative, but their effectiveness in this domain remains unexplored due to the absence of standardized evaluation methodologies. To fill this gap, we introduce CrypFormBench (C.F.B for short), a comprehensive benchmark jointly covering symbolic and computational security to evaluate five core LLM capabilities: interpretation, generation, completion, transformation, and correction. It comprises 700 instances spanning 677 schemes, 7 mainstream formal verifier languages, and 160 security properties. The evaluation of 9 state-of-the-art LLMs reveals that most of them perform well on interpretation and completion, given their code-awareness advantages, but struggle with generation, transformation, and correction. Overall, their performance remains limited, with Claude-3.5 achieving the highest score at 48.7 out of 100. We further provide practical guidance, e.g., few-shot prompting, Pass@K sampling, and lightweight fine-tuning, to mitigate the executability bottleneck and improve tool-usable outputs. Taken together, our benchmark and analyses offer a grounded view of current progress and concrete directions toward reliable LLM-assisted formal cryptographic analysis.

cs.CR↗

arXiv2025

Uncovering and Aligning Anomalous Attention Heads to Defend Against NLP Backdoor Attacks

Backdoor attacks pose a serious threat to the security of large language models (LLMs), causing them to exhibit anomalous behavior under specific trigger conditions. The design of backdoor triggers has evolved from fixed triggers to dynamic or implicit triggers. This increased flexibility in trigger design makes it challenging for defenders to identify their specific forms accurately. Most existing backdoor defense methods are limited to specific types of triggers or rely on an additional clean model for support. To address this issue, we propose a backdoor detection method based on attention similarity, enabling backdoor detection without prior knowledge of the trigger. Our study reveals that models subjected to backdoor attacks exhibit unusually high similarity among attention heads when exposed to triggers. Based on this observation, we propose an attention safety alignment approach combined with head-wise fine-tuning to rectify potentially contaminated attention heads, thereby effectively mitigating the impact of backdoor attacks. Extensive experimental results demonstrate that our method significantly reduces the success rate of backdoor attacks while preserving the model's performance on downstream tasks.

cs.CR↗

arXiv2025

Towards Confidential and Efficient LLM Inference with Dual Privacy Protection

CPU-based trusted execution environments (TEEs) and differential privacy (DP) have gained wide applications for private inference. Due to high inference latency in TEEs, researchers use partition-based approaches that offload linear model components to GPUs. However, dense nonlinear layers of large language models (LLMs) result in significant communication overhead between TEEs and GPUs. DP-based approaches apply random noise to protect data privacy, but this compromises LLM performance and semantic understanding. To overcome the above drawbacks, this paper proposes CMIF, a Confidential and efficient Model Inference Framework. CMIF confidentially deploys the embedding layer in the client-side TEE and subsequent layers on GPU servers. Meanwhile, it optimizes the Report-Noisy-Max mechanism to protect sensitive inputs with a slight decrease in model performance. Extensive experiments on Llama-series models demonstrate that CMIF reduces additional inference overhead in TEEs while preserving user data privacy.

cs.CR↗

arXiv2018

An Extension of the Beurling-Chen-Hadwin-Shen Theorem for Noncommutative Hardy Spaces Associated with Finite von Neumann Algebras

In 2015, Yanni Chen, Don Hadwin and Junhao Shen proved a noncommutative version of Beurling's theorems for a continuous unitarily invariant norm $% α$ on a tracial von Neumann algebra $\left( \mathcal{M},τ\right) $ where $α$ is $\left\Vert \cdot \right\Vert _{1}$-dominating with respect to $τ$. In the paper, we first define a class of norms $% N_{Δ}\left( \mathcal{M},τ\right) $ on $\mathcal{M}$, called determinant, normalized, unitarily invariant continuous norms on $\mathcal{M}$. If $α\in N_{Δ}\left( \mathcal{M},τ\right) $, then there exists a faithful normal tracial state $ρ$ on $\mathcal{M}$ such that $ρ\left( x\right) =τ\left( xg\right) $ for some positive $g\in L^{1}\left( \mathcal{Z},τ\right) $ and the determinant of $g$ is positive. For every $α\in N_{Δ}\left( \mathcal{M},τ\right) $, we study the noncommutative Hardy spaces $% H^{α}\left( \mathcal{M},τ\right) $, then prove that the Chen-Hadwin-Shen theorem holds for $L^{α}\left( \mathcal{M},τ\right) $. The key ingredients in the proof of our result include a factorization theorem and a density theorem for $L^{α}\left( \mathcal{M},ρ\right) $.

math.OA↗

arXiv2016

An Extension of the Chen-Beurling-Helson-Lowdenslager Theorem

Yanni Chen extended the classical Beurling-Helson-Lowdenslager Theorem for Hardy spaces on the unit circle $\mathbb{T}$ defined in terms of continuous gauge norms on $L^{\infty}$ that dominate $\Vert\cdot\Vert_{1}$. We extend Chen's result to a much larger class of continuous gauge norms. A key ingredient is our result that if $α$ is a continuous normalized gauge norm on $L^{\infty}$, then there is a probability measure $λ$, mutually absolutely continuous with respect to Lebesgue measure on $\mathbb{T}$, such that $α\geq c\Vert\cdot\Vert_{1,λ}$ for some $0<c\leq1.$

math.FA↗