arXiv subjects

Sihan Shang

Publications and source records attributed to Sihan Shang.

4 recordsLinked to original sources

arXiv2026

SentAttack: A Sentence-Level Black-Box Adversarial Attack Method for Dense Retrieval Models

Retrieval-Augmented Generation (RAG) systems typically consist of a dense retrieval (DR) model for initial retrieval and a neural ranking model (NRM) for re-ranking.Existing robustness studies in RAG mainly focus on NRMs, while adversarial attacks on DR models are mostly limited to word-level perturbations.For low-ranked target documents that are irrelevant to the query, simple word-level attacks are insufficient to mislead DR models into substantially promoting their rankings.To solve these problems, we propose SentAttack, a sentence-level black-box adversarial attack method for DR models.SentAttack is designed as a two-stage method.In the first stage, SentAttack interacts with the black-box RAG system via iterative retrieval to collect ranked documents and ranking information for training a surrogate DR model.In the second stage, SentAttack uses the surrogate DR model to encode and cluster documents relevant to the target query, yielding multiple cluster centroids.These centroids are concatenated with the target document at the sentence level to form an initial set of adversarial candidates.SentAttack then optimizes these candidates using a query- and centroid-guided objective combined with gradient-guided beam search.Extensive experiments demonstrate that SentAttack outperforms existing adversarial attacks on DR models, with especially strong performance on low-ranked target documents.

cs.IR↗

arXiv2026

Are LLMs Ready to Assist Physicians? PhysAssistBench for Interactive Doctor-Patient-EHR Assistance

The most plausible near-term role of medical LLMs is to assist rather than replace physicians, yet current evaluations often test isolated capabilities: clinical knowledge, EHR system interaction, or patient communication. Physician assistance instead requires coordinating these capabilities within the same interaction, where physicians issue underspecified requests, patients describe symptoms ambiguously, and EHR systems demand precise tool use. We introduce PhysAssistBench, a benchmark for interactive doctor-patient-EHR assistance. Built from real MIMIC-IV cases, PhysAssistBench uses a scalable pipeline to construct agentic patients: interactive, record-grounded agents that turn static EHR records into multi-turn clinical scenarios while preserving clinical factuality. PhysAssistBench provides a curated bilingual evaluation set of 1,296 manually reviewed and physician-validated turns. Experiments with leading LLMs show that current models remain unreliable in this setting, which exposes a key bottleneck for clinical LLMs: reliable assistance requires coordination across knowledge, communication, and systems, not isolated gains in any of them.

cs.CL↗

arXiv2026

ReBeCA: Unveiling Interpretable Behavior Hierarchy behind the Iterative Self-Reflection of Language Models with Causal Analysis

While self-reflection can enhance language model reliability, its underlying mechanisms remain opaque, with existing analyses often yielding correlation-based insights that fail to generalize. To address this, we introduce **ReBeCA** (self-**Re**flection **Be**havior explained through **C**ausal **A**nalysis), a framework for analyzing the interpretable behavioral hierarchy governing the self-reflection outcome. By modeling self-reflection trajectories as causal graphs, ReBeCA selects observed parent candidates and evaluates their stability through a three-stage ICP-based pipeline. In a controlled Qwen3 case study, we establish three critical findings: (1) Behavioral hierarchy: Semantic behaviors of the model influence final self-reflection results hierarchically: directly or indirectly; (2) Causation matters: Generalizability in self-reflection effects is limited to just a few semantic behaviors; (3) More $\neq$ better: The confluence of seemingly positive semantic behaviors, even among direct causal factors, yield no additive gain. ICP-based verification identifies sparse causal parents achieving up to $49.6\%$ structural likelihood gains relative to the dense full-set association baseline across the studied task subsets. A controlled prompt-based behavioral intervention on a novel dataset provides out-of-distribution validation ($p = .013, \eta^2_\mathrm{p} = .071$). The present study focuses on the Qwen3 family under fixed-round Self-Refine.

cs.CL↗

arXiv2024

DataCook: Crafting Anti-Adversarial Examples for Healthcare Data Copyright Protection

In the realm of healthcare, the challenges of copyright protection and unauthorized third-party misuse are increasingly significant. Traditional methods for data copyright protection are applied prior to data distribution, implying that models trained on these data become uncontrollable. This paper introduces a novel approach, named DataCook, designed to safeguard the copyright of healthcare data during the deployment phase. DataCook operates by "cooking" the raw data before distribution, enabling the development of models that perform normally on this processed data. However, during the deployment phase, the original test data must be also "cooked" through DataCook to ensure normal model performance. This process grants copyright holders control over authorization during the deployment phase. The mechanism behind DataCook is by crafting anti-adversarial examples (AntiAdv), which are designed to enhance model confidence, as opposed to standard adversarial examples (Adv) that aim to confuse models. Similar to Adv, AntiAdv introduces imperceptible perturbations, ensuring that the data processed by DataCook remains easily understandable. We conducted extensive experiments on MedMNIST datasets, encompassing both 2D/3D data and the high-resolution variants. The outcomes indicate that DataCook effectively meets its objectives, preventing models trained on AntiAdv from analyzing unauthorized data effectively, without compromising the validity and accuracy of the data in legitimate scenarios. Code and data are available at https://github.com/MedMNIST/DataCook.

cs.AI↗