Searcharxiv⌕ Search

arXiv subjects

Wenqi Li

Publications and source records attributed to Wenqi Li.

At least 19 recordsLinked to original sources
arXiv2026

Do Audio LLMs HEAR Pain? Tracing Acoustic Evidence from Speech to Language Reasoning

Large audio language models (LALMs) can process speech directly, but it remains unclear how well they use the non-lexical acoustic information that speech carries. We study this question through pain, using TAME-Pain, which contains pain-neutral Harvard sentences and spoken pain ratings recorded at different self-reported pain levels. We introduce HEAR, a four-stage diagnostic framework that traces pain-related information from the speech signal to the model's prediction, and apply it to seven closed-source and open-weight LALMs. On the Harvard sentences, where pain can only be inferred from the voice, all models perform at chance level, and same-speaker references or few-shot examples improve balanced accuracy by no more than six points. On the spoken ratings, the same models reach 87.5-99.7% balanced accuracy. Layer-wise probing of open-weight LALMs shows that pain can be decoded from early audio representations (74.1% balanced accuracy for binary prediction and 45.0% for three-class severity), but that this information, along with most pain-associated acoustic cues, decreases in deeper layers. In an acoustic perception test, weaker models cannot distinguish even large changes in these cues, whereas stronger models can but still fail to infer pain. Finally, we recover pain-related acoustic cues and directly reinject them as structured text into the prompt, which improves pairwise pain prediction by up to 32 points. These results indicate that current LALMs encode paralinguistic information in early layers that is largely unavailable when the model produces its answer.

cs.CL↗
arXiv2026

Rigidity of Kähler-Ricci Solitons with Constant Scalar Curvature

Let $(M^{2m},g,f,J)$ be a complete nonsteady gradient Kähler-Ricci soliton satisfying $\mathrm{Ric}+\nabla^2 f=λg$, $λ\neq 0$. We prove that constant scalar curvature forces the soliton to be rigid. More precisely, the universal cover splits holomorphically and isometrically as $N^{2k}\times\mathbb{C}^{m-k}$, where $N^{2k}$ is Kähler-Einstein with $\mathrm{Ric}_{g_N}=λg_N$. In the shrinking case the quotient is trivial; in the normalization $λ=1/2$ one has $R\equiv k$. The Kähler result rests on a Riemannian rigidity criterion. On a complete nonsteady gradient Ricci soliton, if $\mathcal{L}_{\nabla f}\mathrm{Ric}$ is nonnegative or nonpositive everywhere, then it vanishes and the soliton is rigid; no assumption on the scalar curvature is needed. Along the Ricci flow generated by the soliton, this means that a Ricci tensor that is monotone in time is constant in time, and that this forces rigidity. We also give a direct proof that the pinching $0\leq\mathrm{Ric}\leqλg$ forces constant scalar curvature and radial flatness, yielding the rigidity conclusion through the Petersen-Wylie characterization. The proof combines a weighted cutoff argument with a partial Codazzi symmetry for the Ricci endomorphism.

math.DG↗
arXiv2026

A Deterministic Evidence Layer for Vision-Language Autism Screening from Naturalistic Home Video

Autism spectrum disorder (ASD) is diagnosed through specialist observation of a child's social behavior, and access to that expertise is the bottleneck for early identification. Vision-language models (VLMs) describe a child's behavior from video well; the verdict drawn from the description is unstable: at temperature~0, across eight pipeline configurations on one backbone, 16--37\% of clips change their predicted label between repeated runs, and the cause lies in the serving stack. We keep the VLM frozen and move the decision out of the model. Under our grounded perception constraints the VLM writes an event table of timestamped, glossary-labeled events that names the eliciting press and logs counter-evidence; a text-only stage age-calibrates the confidence of each row; a deterministic weight-of-evidence scorer sums it into an evidence total and stratifies it into a risk category, so every decision decomposes into named per-feature contributions and can be re-scored from the saved table. On 43 caregiver-recorded, protocol-free home free-play clips of preschool children, the pipeline reaches AUC $0.851 \pm 0.012$, 86.0\% accuracy, and F$_1$ 71.8 over three runs. It labels 74.4\% of clips correctly in every run (60.5\% for the zero-shot baseline) and flags no typically developing clip in every run (9 of 31 at zero-shot). An ablation on the same backbone attributes the gain to the grounded perception constraints read through the deterministic scorer.

cs.CV↗
arXiv2026

Conditional Accuracy Profiles: Diagnosing LLM Judges across Deployment Conditions

LLM-as-judge is now a standard tool for scalable evaluation, but judge performance is still often summarized by a single accuracy number. This aggregate view hides the deployment conditions under which a judge succeeds or fails. We introduce \textbf{Conditional Accuracy Profiling} (CAP), a post-hoc diagnostic framework that decomposes pairwise LLM-judge accuracy into eight conditions organized into content sensitivity, robustness, and rationale quality. CAP is benchmark-agnostic: it can be applied directly when a benchmark provides the required annotations, approximately through task-subset proxies, or through controlled augmentation when perturbation pairs can be generated. We instantiate CAP on seven LLM judges across six pairwise judging benchmarks, including \textsc{judgerEva-Standard}, a controlled testbed we created to support all eight conditions. CAP exposes profile differences hidden by aggregate accuracy: on \textsc{judgerEva}'s judge-independent Hard-Constructed subset, the two judges most sensitive to omitted qualifications rank in the bottom three of seven by overall accuracy, so omission sensitivity is not predicted by aggregate accuracy. Across benchmarks, Position Robustness shows the strongest rank stability (mean Spearman $\barρ{=}0.87$) but is itself fragile under JudgeBench-Pro adversarial stress, showing the largest mean accuracy drop among the shared conditions, though the dominant degradation channel varies by judge. Condition-level profiles provide a more actionable basis than aggregate accuracy for selecting LLM judges.

cs.LG↗
arXiv2026

Kähler--Ricci Shrinkers with Constant Scalar Curvature Are Rigid

Let $(M^{2m},g,f,J)$ be a complete gradient Kähler--Ricci shrinker satisfying \[ \operatorname{Ric}+\nabla^2 f=\frac12 g. \] We prove that if the scalar curvature is constant, then the soliton is rigid. More precisely, there exists an integer $k\in\{0,\ldots,m\}$ such that \[ R\equiv k \] and \[ (M^{2m},g,J)\cong \bigl(N^{2k}\times\mathbb C^{m-k}, g_N+g_{\mathrm{Euc}},J_N\oplus J_0\bigr), \] where $(N^{2k},g_N,J_N)$ is Kähler--Einstein and \[ \operatorname{Ric}_{g_N}=\frac12 g_N. \] As part of the proof, we establish a Riemannian rigidity criterion: for a complete nonsteady gradient Ricci soliton with constant scalar curvature, the condition $\mathcal{L}_{\nabla f}\operatorname{Ric}=0$ implies radial flatness and hence rigidity.

math.DG↗
arXiv2026

Asymptotic Cohomological Functions and Volumes on Proper Schemes

We study the behavior of asymptotic cohomological functions on proper schemes over a field. We prove a formula for how asymptotic cohomological functions behave under perturbation by an object in the bounded derived category. Using the same method, we also prove that the volume of a line bundle on any proper scheme over a field always exists as a limit.

math.AG↗
arXiv2026

A Proactive Multi-Agent Dialogue Framework for Assessing Social Language Disorder Traits in Autism

Characteristic linguistic behaviors associated with Social Language Disorder (SLD) in autism spectrum disorder, including echoic repetition, pronoun displacement, and stereotyped media quoting, are largely absent from spontaneous conversation and only emerge under specific conversational conditions. In structured clinical assessments, this latency means that questioning strategy selection is a critical yet underappreciated determinant of how much diagnostic information a conversation yields. Whether large language models (LLMs) can be guided to proactively select questioning strategies that systematically surface these latent traits remains largely unexplored. Here we present TPA (Think, Plan, Ask), a proactive multi-agent dialogue framework applied to the language assessment component of the Autism Diagnostic Observation Schedule Module 4 (ADOS-2), in which a doctor agent explicitly reasons about which traits remain unobserved before selecting a clinically grounded strategy and generating a targeted question. A patient agent grounded in real ADOS-2 clinical data enables reproducible evaluation without real patient participation, validated across three independent experiments confirming adequate fidelity to real patient language. Evaluated on 484 episodes from 35 patients, TPA outperforms six competitive dialogue planning baselines across all primary metrics, achieving 82.1% SLD trait coverage, 16.6% higher than automated replay of real clinical dialogues conducted by trained clinicians (65.5%), with substantially greater per-turn diagnostic efficiency (AUCC: 0.628 vs. 0.458, absolute gain +0.170). These results demonstrate that proactive questioning strategy selection substantially improves the efficiency of automated SLD trait assessment, with direct implications for scalable AI-assisted clinical screening.

cs.CL↗
arXiv2026

Ability Transfer and Recovery via Modularized Parameters Localization

Large language models can be continually pre-trained or fine-tuned to improve performance in specific domains, languages, or skills, but this specialization often degrades other capabilities and may cause catastrophic forgetting. We investigate how abilities are distributed within LLM parameters by analyzing module activations under domain- and language-specific inputs for closely related models. Across layers and modules, we find that ability-related activations are highly concentrated in a small set of channels (typically <5\%), and these channels are largely disentangled with good sufficiency and stability. Building on these observations, we propose ACT (Activation-Guided Channel-wise Ability Transfer), which localizes ability-relevant channels via activation differences and selectively transfers only the corresponding parameters, followed by lightweight fine-tuning for compatibility. Experiments on multilingual mathematical and scientific reasoning show that ACT can recover forgotten abilities while preserving retained skills. It can also merge multiple specialized models to integrate several abilities into a single model with minimal interference. Our code and data will be publicly released.

cs.CL↗
arXiv2025

Improving Data Reusability in Interactive Information Retrieval: Insights from the Community

In this study, we conducted semi-structured interviews with 21 IIR researchers to investigate their data reuse practices. This study aims to expand upon current findings by exploring IIR researchers' information-obtaining behaviors regarding data reuse. We identified the information about shared data characteristics that IIR researchers need when evaluating data reusability, as well as the sources they typically consult to obtain this information. We consider this work to be an initial step toward revealing IIR researchers' data reuse practices and identifying what the community needs to do to promote data reuse. We hope that this study, as well as future research, will inspire more individuals to contribute to ongoing efforts aimed at designing standards, infrastructures, and policies, as well as fostering a sustainable culture of data sharing and reuse in this field.

cs.IR↗
arXiv2025

Collaborative Data Behaviors in Digital Humanities Research Teams

The development of digital humanities necessitates scholars to adopt more data-intensive methods and engage in multidisciplinary collaborations. Understanding their collaborative data behaviors becomes essential for providing more curated data, tailored tools, and a collaborative research environment. This study explores how interdisciplinary researchers collaborate on data activities by conducting focus group interviews with 19 digital humanities research groups. Through inductive coding, the study identified seven primary and supportive data activities and found that different collaborative modes are adopted in various data activities. The collaborative modes include humanities-driven, technically-driven, and balanced, depending on how team members naturally adjusted their responsibilities based on their expertise. These findings establish a preliminary framework for examining collaborative data behavior and interdisciplinary collaboration in digital humanities.

cs.DL↗
arXiv2025

VILA-M3: Enhancing Vision-Language Models with Medical Expert Knowledge

Generalist vision language models (VLMs) have made significant strides in computer vision, but they fall short in specialized fields like healthcare, where expert knowledge is essential. In traditional computer vision tasks, creative or approximate answers may be acceptable, but in healthcare, precision is paramount.Current large multimodal models like Gemini and GPT-4o are insufficient for medical tasks due to their reliance on memorized internet knowledge rather than the nuanced expertise required in healthcare. VLMs are usually trained in three stages: vision pre-training, vision-language pre-training, and instruction fine-tuning (IFT). IFT has been typically applied using a mixture of generic and healthcare data. In contrast, we propose that for medical VLMs, a fourth stage of specialized IFT is necessary, which focuses on medical data and includes information from domain expert models. Domain expert models developed for medical use are crucial because they are specifically trained for certain clinical tasks, e.g. to detect tumors and classify abnormalities through segmentation and classification, which learn fine-grained features of medical data$-$features that are often too intricate for a VLM to capture effectively especially in radiology. This paper introduces a new framework, VILA-M3, for medical VLMs that utilizes domain knowledge via expert models. Through our experiments, we show an improved state-of-the-art (SOTA) performance with an average improvement of ~9% over the prior SOTA model Med-Gemini and ~6% over models trained on the specific tasks. Our approach emphasizes the importance of domain expertise in creating precise, reliable VLMs for medical applications.

cs.CV↗
arXiv2025

Beyond Explicit and Implicit: How Users Provide Feedback to Shape Personalized Recommendation Content

As personalized recommendation algorithms become integral to social media platforms, users are increasingly aware of their ability to influence recommendation content. However, limited research has explored how users provide feedback through their behaviors and platform mechanisms to shape the recommendation content. We conducted semi-structured interviews with 34 active users of algorithmic-driven social media platforms (e.g., Xiaohongshu, Douyin). In addition to explicit and implicit feedback, this study introduced intentional implicit feedback, highlighting the actions users intentionally took to refine recommendation content through perceived feedback mechanisms. Additionally, choices of feedback behaviors were found to align with specific purposes. Explicit feedback was primarily used for feed customization, while unintentional implicit feedback was more linked to content consumption. Intentional implicit feedback was employed for multiple purposes, particularly in increasing content diversity and improving recommendation relevance. This work underscores the user intention dimension in the explicit-implicit feedback dichotomy and offers insights for designing personalized recommendation feedback that better responds to users' needs.

cs.HC↗
arXiv2025

An Exceptional Dataset For Rare Pancreatic Tumor Segmentation

Pancreatic NEuroendocrine Tumors (pNETs) are very rare endocrine neoplasms that account for less than 5% of all pancreatic malignancies, with an incidence of only 1-1.5 cases per 100,000. Early detection of pNETs is critical for improving patient survival, but the rarity of pNETs makes segmenting them from CT a very challenging problem. So far, there has not been a dataset specifically for pNETs available to researchers. To address this issue, we propose a pNETs dataset, a well-annotated Contrast-Enhanced Computed Tomography (CECT) dataset focused exclusively on Pancreatic Neuroendocrine Tumors, containing data from 469 patients. This is the first dataset solely dedicated to pNETs, distinguishing it from previous collections. Additionally, we provide the baseline detection networks with a new slice-wise weight loss function designed for the UNet-based model, improving the overall pNET segmentation performance. We hope that our dataset can enhance the understanding and diagnosis of pNET Tumors within the medical community, facilitate the development of more accurate diagnostic tools, and ultimately improve patient outcomes and advance the field of oncology.

cs.CV↗
arXiv2024

Exploiting ChatGPT for Diagnosing Autism-Associated Language Disorders and Identifying Distinct Features

Diagnosing language disorders associated with autism is a complex challenge, often hampered by the subjective nature and variability of traditional assessment methods. Traditional diagnostic methods not only require intensive human effort but also often result in delayed interventions due to their lack of speed and precision. In this study, we explored the application of ChatGPT, a large language model, to overcome these obstacles by enhancing sensitivity and profiling linguistic features for autism diagnosis. This research utilizes ChatGPT natural language processing capabilities to simplify and improve the diagnostic process, focusing on identifying autism related language patterns. Specifically, we compared ChatGPT performance with that of conventional supervised learning models, including BERT, a model acclaimed for its effectiveness in various natural language processing tasks. We showed that ChatGPT substantially outperformed these models, achieving over 10% improvement in both sensitivity and positive predictive value, in a zero shot learning configuration. The findings underscore the model potential as a diagnostic tool, combining accuracy and applicability. We identified ten key features of autism associated language disorders across scenarios. Features such as echolalia, pronoun reversal, and atypical language usage play a critical role in diagnosing ASD and informing tailored treatment plans. Together, our findings advocate for adopting sophisticated AI tools like ChatGPT in clinical settings to assess and diagnose developmental disorders. Our approach promises enhanced diagnostic precision and supports personalized medicine, potentially transforming the evaluation landscape for autism and similar neurological conditions.

cs.CL↗
arXiv2024

The Landscape of Data Reuse in Interactive Information Retrieval: Motivations, Sources, and Evaluation of Reusability

Sharing and reusing research data can effectively reduce redundant efforts in data collection and curation, especially for small labs and research teams conducting human-centered system research, and enhance the replicability of evaluation experiments. Building a sustainable data reuse process and culture relies on frameworks that encompass policies, standards, roles, and responsibilities, all of which must address the diverse needs of data providers, curators, and reusers. To advance the knowledge and accumulate empirical understandings on data reuse, this study investigated the data reuse practices of experienced researchers from the area of Interactive Information Retrieval (IIR) studies, where data reuse has been strongly advocated but still remains a challenge. To enhance the knowledge on data reuse behavior and reusability assessment strategies within IIR community, we conducted 21 semi-structured in-depth interviews with IIR researchers from varying demographic backgrounds, institutions, and stages of careers on their motivations, experiences, and concerns over data reuse. We uncovered the reasons, strategies of reusability assessments, and challenges faced by data reusers within the field of IIR as they attempt to reuse researcher data in their studies. The empirical finding improves our understanding of researchers' motivations for reusing data, their approaches to discovering reusable research data, as well as their concerns and criteria for assessing data reusability, and also enriches the on-going discussions on evaluating user-generated data and research resources and promoting community-level data reuse culture and standards.

cs.IR↗
arXiv2024

VISTA3D: A Unified Segmentation Foundation Model For 3D Medical Imaging

Foundation models for interactive segmentation in 2D natural images and videos have sparked significant interest in building 3D foundation models for medical imaging. However, the domain gaps and clinical use cases for 3D medical imaging require a dedicated model that diverges from existing 2D solutions. Specifically, such foundation models should support a full workflow that can actually reduce human effort. Treating 3D medical images as sequences of 2D slices and reusing interactive 2D foundation models seems straightforward, but 2D annotation is too time-consuming for 3D tasks. Moreover, for large cohort analysis, it's the highly accurate automatic segmentation models that reduce the most human effort. However, these models lack support for interactive corrections and lack zero-shot ability for novel structures, which is a key feature of "foundation". While reusing pre-trained 2D backbones in 3D enhances zero-shot potential, their performance on complex 3D structures still lags behind leading 3D models. To address these issues, we present VISTA3D, Versatile Imaging SegmenTation and Annotation model, that targets to solve all these challenges and requirements with one unified foundation model. VISTA3D is built on top of the well-established 3D segmentation pipeline, and it is the first model to achieve state-of-the-art performance in both 3D automatic (supporting 127 classes) and 3D interactive segmentation, even when compared with top 3D expert models on large and diverse benchmarks. Additionally, VISTA3D's 3D interactive design allows efficient human correction, and a novel 3D supervoxel method that distills 2D pretrained backbones grants VISTA3D top 3D zero-shot performance. We believe the model, recipe, and insights represent a promising step towards a clinically useful 3D foundation model. Code and weights are publicly available at https://github.com/Project-MONAI/VISTA.

cs.CV↗
arXiv2024

Video-based Analysis Reveals Atypical Social Gaze in People with Autism Spectrum Disorder

In this study, we present a quantitative and comprehensive analysis of social gaze in people with autism spectrum disorder (ASD). Diverging from traditional first-person camera perspectives based on eye-tracking technologies, this study utilizes a third-person perspective database from the Autism Diagnostic Observation Schedule, 2nd Edition (ADOS-2) interview videos, encompassing ASD participants and neurotypical individuals as a reference group. Employing computational models, we extracted and processed gaze-related features from the videos of both participants and examiners. The experimental samples were divided into three groups based on the presence of social gaze abnormalities and ASD diagnosis. This study quantitatively analyzed four gaze features: gaze engagement, gaze variance, gaze density map, and gaze diversion frequency. Furthermore, we developed a classifier trained on these features to identify gaze abnormalities in ASD participants. Together, we demonstrated the effectiveness of analyzing social gaze in people with ASD in naturalistic settings, showcasing the potential of third-person video perspectives in enhancing ASD diagnosis through gaze analysis.

q-bio.NC↗