MLMing Lu

Ming Lu

*This is an algorithmically generated author record based on publicly available information.

Ming Lu has a total of 19 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Top Cited Papers

    Loki’s Dance of Illusions: A Comprehensive Survey of Hallucination in Large Language Models

    by Ming Lu, Chaozhuo Li, Pengbo Wang, Chenxu Wang, Litian Zhang, Zheng Liu, Qiwei Ye, Yi Hua, Yushan Cai, Yuanbo Xu, et al.

    Edgar Allan Poe noted, “Truth often lurks in the shadow of error,” highlighting the deep complexity intrinsic to the interplay between truth and falsehood, notably under conditions of cognitive and informational asymmetry. This dynamic is strikingly evident in large language models (LLMs). Despite their impressive linguistic generation capabilities, LLMs sometimes produce information that appears factually accurate but is, in reality, fabricated, an issue often referred to as “hallucinations.” The prevalence of these hallucinations can mislead users, affecting their judgments and decisions. In sectors such as finance, law, and healthcare, such misinformation risks causing substantial economic losses, legal disputes, and health risks, with wide-ranging consequences. In our research, we have methodically categorized and analyzed the causes, detection methods, and solutions related to LLM hallucinations. Our efforts specifically focus on elucidating the root causes of hallucinations and evaluating current strategies, thereby laying the groundwork for developing robust mitigation techniques.


  • ParallelGSE: Efficient and Secure Shortest Path Search on Encrypted Graphs

    by Qing Fan, Weixiao Wang, Chuan Zhang, Zhitao Guan, Yong Xie, Ming Lu, Liehuang Zhu

    Graph searchable encryption (GSE) for shortest path queries allows users to discover the closest connection between two individuals in encrypted social network graphs, while safeguarding both social data security and the privacy of user queries. The static GSE is commonly favored for its higher efficiency, while it faces significant challenges in resisting query recovery attacks. In contrast, the dynamic GSE is better suited for changing real world, but enhancing its search efficiency remains a formidable challenge. In this paper, we propose the ParallelGSE, a flexible graph partitioning and parallel search framework that supports both dynamic and static encrypted social graphs, striking a balance between query privacy and practical implementation. ParallelGSE partitions the original graph into multiple subgraphs, distributes them across several semi-honest servers, and performs searches in parallel without relying on a trusted server. Formal security reduction demonstrates ParallelGSE is resistant to isomorphic query recovery attacks. Simulated experiments on seven real-world graph datasets validate rationality of graph partition and improvements of storage, computation and communication costs. Especially, compared to the state-of-the-art (SOTA) static PathGES (ACM CCS 2024) and dynamic GraphShield (IEEE TKDE 2022), the search efficiency of static ParallelGSE can be 3 orders of magnitude faster than that of PathGSE, while dynamic ParallelGSE achieves over $20\times$ greater search efficiency than GraphShield.