arXiv subjects

Mingxuan Zheng

Publications and source records attributed to Mingxuan Zheng.

4 recordsLinked to original sources

arXiv2026

LexAgentHallu: A Hierarchical Benchmark for Profiling Hallucinations in Legal Agents

As large language models are increasingly deployed as tool-augmented legal agents, they introduce agentic hallucinations where tool-call and reasoning errors cascade into fabricated holdings and miscited authority. However, existing legal benchmarks evaluate only single-turn QA with outcome-level metrics, while agentic hallucination benchmarks lack legal-specific diagnostic capability. Neither answers to what extent and how a legal agent hallucinates along its trajectory. To address these limitations, we introduce LexAgentHallu, a legal agentic hallucination benchmark designed to evaluate to what extent and how legal agents fail along multi-step trajectories. Built through a four-stage expert-in-the-loop pipeline, LexAgentHallu contains 3414 instances across 17 legal categories and 6 task types. Each instance is annotated under a dual-layer hallucination taxonomy of 7 high-level categories and 27 fine-grained subclasses, covering both substantive errors and agent-procedural failures. We further design fine-grained metrics that quantify to what extent and localize how each failure occurs along an agent's execution path. Our evaluation across 18 proprietary and open-source agents uncovers a Right-Answer-Wrong-Reason effect and reveals that hallucination subclasses cluster rather than scatter, forming distinct agentic framework, legal task, and category profiles. These findings, invisible to outcome-level evaluation, validate the diagnostic power of LexAgentHallu for evaluating agentic hallucination in law.

cs.AI↗

arXiv2026

SkillProx: Self-Evolving Agent Skills via Proximal Textual Gradient Descent

LLM agents increasingly adapt to recurring tasks by accumulating procedural knowledge in skills. These skills are lightweight, reusable textual artifacts that are loaded into the agent's context without weight updates. Recent methods refine skills through iterative task execution, failure diagnosis, and trajectory-guided text-space updates. However, existing frameworks lack explicit diagnosis--outcome feedback and treat deletion as a generic edit operation rather than a dedicated mechanism for consolidating accumulated knowledge. We introduce SkillProx, a proximal-gradient-inspired forward--backward framework that couples closed-loop diagnostic evolution with utility-aware proximal refinement. Motivated by a composite objective balancing task loss and skill complexity, the forward stage re-executes diagnosis-driven edits on the same task batch, rolls back regressions, and feeds measured outcomes into subsequent diagnoses. The backward stage decomposes the resulting skill into auditable knowledge units, estimates their contributions using a frozen leave-one-out utility audit, and applies validation-gated consolidation, demotion, or removal. Experiments on in-distribution and out-of-distribution benchmarks across multiple backbone LLMs show that SkillProx improves average accuracy by 3.0 percentage points over the strongest gradient-based baseline. Component ablations demonstrate the complementary effects of closed-loop diagnosis and proximal refinement.

cs.AI↗

arXiv2025

Structured Event Representation and Stock Return Predictability

We find that event features extracted by large language models (LLMs) are effective for text-based stock return prediction. Using a pre-trained LLM to extract event features from news articles, we propose a novel deep learning model based on structured event representation (SER) and attention mechanisms to predict stock returns in the cross-section. Our SER-based model provides superior performance compared with other existing text-driven models to forecast stock returns out of sample and offers highly interpretable feature structures to examine the mechanisms underlying the stock return predictability. We further provide various implications based on SER and highlight the crucial benefit of structured model inputs in stock return predictability.

econ.GN↗

arXiv2024

Sensing-aided Near-Field Secure Communications with Mobile Eavesdroppers

The additional degree of freedom (DoF) in the distance domain of near-field communication offers new opportunities for physical layer security (PLS) design. However, existing works mainly consider static eavesdroppers, and the related study with mobile eavesdroppers is still in its infancy due to the difficulty in obtaining the channel state information (CSI) of the eavesdropper. To this end, we propose to leverage the sensing capability of integrated sensing and communication (ISAC) systems to assist PLS design. To comprehensively study the dynamic behaviors of the system, we propose a Pareto optimization framework, where a multi-objective optimization problem (MOOP) is formulated to simultaneously optimize three key performance metrics: power consumption, number of securely served users, and tracking performance, while guaranteeing the achievable rate of the users with a given leakage rate constraint. A globally optimal design based on the generalized Benders decomposition (GBD) method is proposed to achieve the Pareto optimal solutions. To reduce the computational complexity, we further design a low-complexity algorithm based on zero-forcing (ZF) beamforming and successive convex approximation (SCA). Simulation results validate the effectiveness of the proposed algorithms and reveal the intrinsic trade-offs between the three performance metrics. It is observed that near-field communication offers a favorable beam diffraction effect for PLS, where the energy of the information signal is nulled around the eavesdropper and focused on the users.

eess.SP↗