arXiv subjects

Zhiyuan Fu

Publications and source records attributed to Zhiyuan Fu.

4 recordsLinked to original sources

arXiv2026

AgentHijack: Visual Patch Attacks on Multimodal Computer-Use Agents

This paper presents an end-to-end evaluation framework for image-triggered command injection against computer-use agents (CUAs). The goal is to test whether a local visual patch can induce verifiable environmental consequences along the full chain of screenshot input, VLM generation, action parsing, and environment execution. We train and deploy patches on author-controlled GitHub Pages pages and a locally deployed CSDN clone, and evaluate them in real environments across five open-source or publicly available GUI-agent or vision-language-model (VLM) backends. Our experiment aggregates 600 instance-level online cases, with T-ASR, TAPR, and E2E-ASR reaching 84.5%, 47.0%, and 20.3%, respectively. Trajectory analysis further shows that in some successful cases the agent first executes a malicious terminal command and then continues the original benign task. These results indicate that optimized local visual signals can affect not only VLM outputs but also propagate through the execution pipeline of open CUAs and create real environmental risk.

cs.CR↗

arXiv2026

FDLLM: A Dedicated Detector for Black-Box LLMs Fingerprinting

Large Language Models (LLMs) are rapidly transforming the landscape of digital content creation. However, the prevalent black-box Application Programming Interface (API) access to many LLMs introduces significant challenges in accountability, governance, and security. LLM fingerprinting, which aims to identify the source model by analyzing statistical and stylistic features of generated text, offers a potential solution. Current progress in this area is hindered by a lack of dedicated datasets and the need for efficient, practical methods that are robust against adversarial manipulations. To address these challenges, we introduce FD-Dataset, a comprehensive bilingual fingerprinting benchmark comprising 90,000 text samples from 20 famous proprietary and open-source LLMs. Furthermore, we present FDLLM, a novel fingerprinting method that leverages parameter-efficient Low-Rank Adaptation (LoRA) to fine-tune a foundation model. This approach enables LoRA to extract deep, persistent features that characterize each source LLM. Through our analysis, we find that LoRA adaptation promotes the aggregation of outputs from the same LLM in representation space while enhancing the separation between different LLMs. This mechanism explains why LoRA proves particularly effective for LLM fingerprinting. Extensive empirical evaluations on FD-Dataset demonstrate FDLLM's superiority, achieving a Macro F1 score 22.1% higher than the strongest baseline. FDLLM also exhibits strong generalization to newly released models, achieving an average accuracy of 95% on unseen models. Notably, FDLLM remains consistently robust under various adversarial attacks, including polishing, translation, and synonym substitution. Experimental results show that FDLLM reduces the average attack success rate from 49.2% (LM-D) to 23.9%.

cs.CR↗

arXiv2024

BMO on Weighted Bergman Spaces over Tubular Domains

In this paper, we characterize Bounded Mean Oscillation (BMO) and establish their connection with Hankel operators on weighted Bergman spaces over tubular domains. By utilizing the space BMO, we provide a new characterization of Bloch spaces on tubular domains. Next, we define a modified projection operator and prove its boundedness. Furthermore, we introduce differential operators and demonstrate that these operators belong to Lebesgue spaces on tubular domains. Finally, we establish an integral representation for Bergman functions using these differential operators.

math.CV↗

arXiv2020

Projecting the optimal control strategy on invasive plants combining effects of herbivores and native plants resistance

Understanding how to limit biological invasion is critical, especially in the context of accelerating anthropogenic ecological changes. Although biological invasion success could be explained by the lack of natural enemies in new regions, recent studies have revealed that resident herbivores often do have a substantial effect on both native and invasive plants. Very few studies have included consideration of native plant resistance while estimating methods of controlling invasion; hence, it is unclear to what extent the interactive effects of controlling approaches and native plants' resistance could slow down or even inhibit biological invasion. We developed a spatial modeling framework, using a paired logistic equation model, with considerations of the dispersal processes, to capture the dynamics change of native and invasive plants under various strategies of control. We found that when biocontrol agents could have a strong effect on invasive plant, that could almost completely limit the invasion, together with a high native plant resistance. However, a high application frequency is needed make an efficient impact, whereas, a low frequency treatment leads to nearly the same outcome as the no treatment case. Lastly, we showed that evenly controlling a larger area with a weaker effect still lead to a better outcome than focusing on small patches with a stronger effect. Overall, this study has some management implications, such as how to determine the optimal allocation strategy.

q-bio.PE↗