arXiv subjects

Chenguang Wang

Publications and source records attributed to Chenguang Wang.

At least 55 records · Page 3Linked to original sources

arXiv2026

Source localization realizes single frame super-resolution for fluorescence imaging

Existing super-resolution microscopy is often constrained by inherent trade-offs between resolution, acquisition speed, phototoxicity, and hardware complexity. Computational post-processing approaches offer a promising alternative, but they typically suffer from linearity distortion, high computational cost, reliance on pre-training data, or reconstruction artifacts. Here, we present Source Localization (SoLo), a novel single-frame super-resolution algorithm for fluorescence imaging without these limitations. Built on the principle of inferring fluorescent source positions via sampling-detection strategy, SoLo achieves non-iterative, parallelizable computation, enabling real-time live-cell imaging with high spatiotemporal resolution. The intensity linearity preservation of SoLo makes it compatible with quantitative analysis such as calcium imaging and fluorescence resonance energy transfer. We further extended this framework to 3D-SoLo for volumetric imaging and nonlinear SoLo (NL-SoLo) for high-density fluorescence fluctuation imaging. With its ease of parameter tuning and compatibility with existing imaging systems, SoLo offers an accessible solution for ordinary labs, enabling diverse biomedical imaging applications.

physics.optics↗

arXiv2026

RepIt: Steering Language Models with Concept-Specific Refusal Vectors

Current safety evaluations of language models rely on benchmark-based assessments that may miss localized vulnerabilities. We present RepIt, a simple and data-efficient framework for isolating concept-specific representations in LM activations. While existing steering methods already achieve high attack success rates through broad interventions, RepIt enables a more concerning capability: selective suppression of refusal on targeted concepts while preserving refusal elsewhere. Across five frontier LMs, RepIt produces evaluation-evading model organisms with semantic backdoors, answering questions related to weapons of mass destruction while still scoring as safe on standard benchmarks. We find the edit of the steering vector localizes to just 100-200 residual dimensions, and robust concept vectors can be extracted from as few as a dozen examples on a single RTX A6000, highlighting how targeted, hard-to-detect modifications can exploit evaluation blind spots with minimal resources. Through demonstrating precise concept disentanglement, this work exposes vulnerabilities in current safety evaluation practices and demonstrates a need for more comprehensive, representation aware assessments.

cs.AI↗

arXiv2026

A Framework for Formalizing LLM Agent Security

Security in LLM agents is inherently contextual. For example, the same action taken by an agent may represent legitimate behavior or a security violation depending on whose instruction led to the action, what objective is being pursued, and whether the action serves that objective. However, existing definitions of security attacks against LLM agents often fail to capture this contextual nature. As a result, defenses face a fundamental utility-security tradeoff: applying defenses uniformly across all contexts can lead to significant utility loss, while applying defenses in insufficient or inappropriate contexts can result in security vulnerabilities. In this work, we present a framework that systematizes existing attacks and defenses from the perspective of contextual security. To this end, we propose four security properties that capture contextual security for LLM agents: task alignment (pursuing authorized objectives), action alignment (individual actions serving those objectives), source authorization (executing commands from authenticated sources), and data isolation (ensuring information flows respect privilege boundaries). We further introduce a set of oracle functions that enable verification of whether these security properties are violated as an agent executes a user task. Using this framework, we reformalize existing attacks, such as indirect prompt injection, direct prompt injection, jailbreak, task drift, and memory poisoning, as violations of one or more security properties, thereby providing precise and contextual definitions of these attacks. Similarly, we reformalize defenses as mechanisms that strengthen oracle functions or perform security property checks. Finally, we discuss several important future research directions enabled by our framework.

cs.CR↗

arXiv2026

From Image Generation to Infrastructure Design: a Multi-agent Pipeline for Street Design Generation

Realistic visual renderings of street-design scenarios are essential for public engagement in active transportation planning. Traditional approaches are labor-intensive, hindering collective deliberation and collaborative decision-making. While AI-assisted generative design shows transformative potential by enabling rapid creation of design scenarios, existing generative approaches typically require large amounts of domain-specific training data and struggle to enable precise spatial variations of design/configuration in complex street-view scenes. We introduce a multi-agent system that edits and redesigns bicycle facilities directly on real-world street-view imagery. The framework integrates lane localization, prompt optimization, design generation, and automated evaluation to synthesize realistic, contextually appropriate designs. Experiments across diverse urban scenarios demonstrate that the system can adapt to varying road geometries and environmental conditions, consistently yielding visually coherent and instruction-compliant results. This work establishes a foundation for applying multi-agent pipelines to transportation infrastructure planning and facility design.

cs.AI↗

arXiv2026

Are Your Generated Instances Truly Useful? GenBench-MILP: A Benchmark Suite for MILP Instance Generation

The proliferation of machine learning-based methods for Mixed-Integer Linear Programming (MILP) instance generation has surged, driven by the need for diverse training datasets. However, a critical question remains: Are these generated instances truly useful and realistic? Current evaluation protocols often rely on superficial structural metrics or simple solvability checks, which frequently fail to capture the true computational complexity of real-world problems. To bridge this gap, we introduce GenBench-MILP, a comprehensive benchmark suite designed for the standardized and objective evaluation of MILP generators. Our framework assesses instance quality across four key dimensions: mathematical validity, structural similarity, computational hardness, and utility in downstream tasks. A distinctive innovation of GenBench-MILP is the analysis of solver-internal features -- including root node gaps, heuristic success rates, and cut plane usage. By treating the solver's dynamic behavior as an expert assessment, we reveal nuanced computational discrepancies that static graph features miss. Our experiments on instance generative models demonstrate that instances with high structural similarity scores can still exhibit drastically divergent solver interactions and difficulty levels. By providing this multifaceted evaluation toolkit, GenBench-MILP aims to facilitate rigorous comparisons and guide the development of high-fidelity instance generators.

cs.LG↗

arXiv2026

ALPBench: A Benchmark for Attribution-level Long-term Personal Behavior Understanding

Recent advances in large language models have highlighted their potential for personalized recommendation, where accurately capturing user preferences remains a key challenge. Leveraging their strong reasoning and generalization capabilities, LLMs offer new opportunities for modeling long-term user behavior. To systematically evaluate this, we introduce ALPBench, a Benchmark for Attribution-level Long-term Personal Behavior Understanding. Unlike item-focused benchmarks, ALPBench predicts user-interested attribute combinations, enabling ground-truth evaluation even for newly introduced items. It models preferences from long-term historical behaviors rather than users' explicitly expressed requests, better reflecting enduring interests. User histories are represented as natural language sequences, allowing interpretable, reasoning-based personalization. ALPBench enables fine-grained evaluation of personalization by focusing on the prediction of attribute combinations task that remains highly challenging for current LLMs due to the need to capture complex interactions among multiple attributes and reason over long-term user behavior sequences.

cs.IR↗

arXiv2026

Persona-aware and Explainable Bikeability Assessment: A Vision-Language Model Approach

Bikeability assessment is essential for advancing sustainable urban transportation and creating cyclist-friendly cities, and it requires incorporating users' perceptions of safety and comfort. Yet existing perception-based bikeability assessment approaches face key limitations in capturing the complexity of road environments and adequately accounting for heterogeneity in subjective user perceptions. This paper proposes a persona-aware Vision-Language Model framework for bikeability assessment with three novel contributions: (i) theory-grounded persona conditioning based on established cyclist typology that generates persona-specific explanations via chain-of-thought reasoning; (ii) multi-granularity supervised fine-tuning that combines scarce expert-annotated reasoning with abundant user ratings for joint prediction and explainable assessment; and (iii) AI-enabled data augmentation that creates controlled paired data to isolate infrastructure variable impacts. To test and validate this framework, we developed a panoramic image-based crowdsourcing system and collected 12,400 persona-conditioned assessments from 427 cyclists. Experiment results show that the proposed framework offers competitive bikeability rating prediction while uniquely enabling explainable factor attribution.

cs.CL↗

arXiv2025

LLM CHESS: Benchmarking Reasoning and Instruction-Following in LLMs through Chess

We introduce LLM CHESS, an evaluation framework designed to probe the generalization of reasoning and instruction-following abilities in large language models (LLMs) through extended agentic interaction in the domain of chess. We rank over 50 open and closed source models by playing against a random opponent using a range of behavioral metrics, including win and loss rates, move quality, move legality, hallucinated actions, and game duration. For a subset of top reasoning models, we derive an Elo estimate by playing against a chess engine with variably configured skill, which allows for comparisons between models in an easily understandable way. Despite the simplicity of the instruction-following task and the weakness of the opponent, many state-of-the-art models struggle to complete games or achieve consistent wins. Similar to other benchmarks on complex reasoning tasks, our experiments reveal a clear separation between reasoning and non-reasoning models. However, unlike existing static benchmarks, the stochastic and dynamic nature of LLM CHESS uniquely reduces overfitting and memorization while preventing benchmark saturation, proving difficult even for top reasoning models. To support future work on evaluating reasoning and instruction-following in LLMs, we release our experimental framework, a public leaderboard, and a dataset of associated games.

cs.AI↗

arXiv2025

VMDT: Decoding the Trustworthiness of Video Foundation Models

As foundation models become more sophisticated, ensuring their trustworthiness becomes increasingly critical; yet, unlike text and image, the video modality still lacks comprehensive trustworthiness benchmarks. We introduce VMDT (Video-Modal DecodingTrust), the first unified platform for evaluating text-to-video (T2V) and video-to-text (V2T) models across five key trustworthiness dimensions: safety, hallucination, fairness, privacy, and adversarial robustness. Through our extensive evaluation of 7 T2V models and 19 V2T models using VMDT, we uncover several significant insights. For instance, all open-source T2V models evaluated fail to recognize harmful queries and often generate harmful videos, while exhibiting higher levels of unfairness compared to image modality models. In V2T models, unfairness and privacy risks rise with scale, whereas hallucination and adversarial robustness improve -- though overall performance remains low. Uniquely, safety shows no correlation with model size, implying that factors other than scale govern current safety levels. Our findings highlight the urgent need for developing more robust and trustworthy video foundation models, and VMDT provides a systematic framework for measuring and tracking progress toward this goal. The code is available at https://sunblaze-ucb.github.io/VMDT-page/.

cs.CV↗

arXiv2025

A Flow-Based Model for Conditional and Probabilistic Electricity Consumption Profile Generation and Prediction

Residential Load Profile (RLP) generation and prediction are critical for the operation and planning of distribution networks, especially as diverse low-carbon technologies (e.g., photovoltaic and electric vehicles) are increasingly adopted. This paper introduces a novel flow-based generative model, termed Full Convolutional Profile Flow (FCPFlow), which is uniquely designed for both conditional and unconditional RLP generation, and for probabilistic load forecasting. By introducing two new layers--the invertible linear layer and the invertible normalization layer--the proposed FCPFlow architecture shows three main advantages compared to traditional statistical and contemporary deep generative models: 1) it is well-suited for RLP generation under continuous conditions, such as varying weather and annual electricity consumption, 2) it demonstrates superior scalability in different datasets compared to traditional statistical models, and 3) it also demonstrates better modeling capabilities in capturing the complex correlation of RLPs compared with deep generative models.

cs.LG↗

arXiv2025

Budget-aware Test-time Scaling via Discriminative Verification

Test-time scaling is a powerful strategy for boosting the performance of large language models on complex reasoning tasks. While state-of-the-art approaches often employ generative verifiers to select the best solution from a pool of candidates, this method incurs prohibitive computational costs, limiting its practicality. In this work, we shift the focus to a more budget-aware paradigm: discriminative verification. We conduct a thorough empirical analysis and demonstrate that while discriminative verifiers may underperform in isolation, combining them with self-consistency in a hybrid approach creates a powerful and efficient test-time scaling mechanism. Notably, under a fixed compute budget, this hybrid approach surpasses state-of-the-art generative verification by a significant margin: achieving up to 15.3\% higher accuracy on AIME2025. Our findings establish that for practical, real-world applications, budget-aware scaling with discriminative verifiers is not only a "free" upgrade over self-consistency, but also a more effective and efficient alternative to costly generative techniques. Code is available at https://github.com/wang-research-lab/verification.

cs.AI↗

arXiv2025

Predicting Task Performance with Context-aware Scaling Laws

Scaling laws have transformed our understanding of large language models by linking upstream metrics like cross-entropy loss to design factors such as model size, training data, and compute. However, these conventional laws fail to capture downstream task performance, where context plays a critical role. In this work, we propose a straightforward, interpretable framework that jointly models downstream performance as a function of the training compute and the provided context. We empirically validate our framework by fitting it on the observed downstream performance of extended-context variants of Llama-2-7B and Llama-2-13B across 65,500 unique instances spanning three tasks: arithmetic reasoning, common sense reasoning, and machine translation. Our results demonstrate that our framework accurately models in-distribution downstream performance, generalizes across three orders of magnitude in training compute, and reliably extrapolates performance as the amount of context increases. These findings offer valuable insights into the interplay between training compute and context utilization, providing guidance for designing more efficient long-context LLMs for diverse downstream tasks. Our code is available at https://github.com/wang-research-lab/context-scaling.

cs.CL↗

arXiv2025

Towards Visual Text Grounding of Multimodal Large Language Model

Despite the existing evolution of Multimodal Large Language Models (MLLMs), a non-neglectable limitation remains in their struggle with visual text grounding, especially in text-rich images of documents. Document images, such as scanned forms and infographics, highlight critical challenges due to their complex layouts and textual content. However, current benchmarks do not fully address these challenges, as they mostly focus on visual grounding on natural images, rather than text-rich document images. Thus, to bridge this gap, we introduce TRIG, a novel task with a newly designed instruction dataset for benchmarking and improving the Text-Rich Image Grounding capabilities of MLLMs in document question-answering. Specifically, we propose an OCR-LLM-human interaction pipeline to create 800 manually annotated question-answer pairs as a benchmark and a large-scale training set of 90$ synthetic data based on four diverse datasets. A comprehensive evaluation of various MLLMs on our proposed benchmark exposes substantial limitations in their grounding capability on text-rich images. In addition, we propose two simple and effective TRIG methods based on general instruction tuning and plug-and-play efficient embedding, respectively. By finetuning MLLMs on our synthetic dataset, they promisingly improve spatial reasoning and grounding capabilities.

cs.CV↗

arXiv2025

VisR-Bench: An Empirical Study on Visual Retrieval-Augmented Generation for Multilingual Long Document Understanding

Most organizational data in this world are stored as documents, and visual retrieval plays a crucial role in unlocking the collective intelligence from all these documents. However, existing benchmarks focus on English-only document retrieval or only consider multilingual question-answering on a single-page image. To bridge this gap, we introduce VisR-Bench, a multilingual benchmark designed for question-driven multimodal retrieval in long documents. Our benchmark comprises over 35K high-quality QA pairs across 1.2K documents, enabling fine-grained evaluation of multimodal retrieval. VisR-Bench spans sixteen languages with three question types (figures, text, and tables), offering diverse linguistic and question coverage. Unlike prior datasets, we include queries without explicit answers, preventing models from relying on superficial keyword matching. We evaluate various retrieval models, including text-based methods, multimodal encoders, and MLLMs, providing insights into their strengths and limitations. Our results show that while MLLMs significantly outperform text-based and multimodal encoder models, they still struggle with structured tables and low-resource languages, highlighting key challenges in multilingual visual retrieval.

cs.CV↗

arXiv2025

MusiXQA: Advancing Visual Music Understanding in Multimodal Large Language Models

Multimodal Large Language Models (MLLMs) have achieved remarkable visual reasoning abilities in natural images, text-rich documents, and graphic designs. However, their ability to interpret music sheets remains underexplored. To bridge this gap, we introduce MusiXQA, the first comprehensive dataset for evaluating and advancing MLLMs in music sheet understanding. MusiXQA features high-quality synthetic music sheets generated via MusiXTeX, with structured annotations covering note pitch and duration, chords, clefs, key/time signatures, and text, enabling diverse visual QA tasks. Through extensive evaluations, we reveal significant limitations of current state-of-the-art MLLMs in this domain. Beyond benchmarking, we developed Phi-3-MusiX, an MLLM fine-tuned on our dataset, achieving significant performance gains over GPT-based methods. The proposed dataset and model establish a foundation for future advances in MLLMs for music sheet understanding. Code, data, and model will be released upon acceptance.

cs.CV↗

arXiv2025

Cure Rate Joint Model for Time-to-Event Data and Longitudinal Tumor Burden with Potential Change Points

In non-small cell lung cancer (NSCLC) clinical trials, tumor burden (TB) is a key longitudinal biomarker for assessing treatment effects. Typically, standard-of-care (SOC) therapies and some novel interventions initially decrease TB; however, many patients subsequently experience an increase-indicating disease progression-while others show a continuous decline. In patients with an eventual TB increase, the change point marks the onset of progression and must occur before the time of the event. To capture these distinct dynamics, we propose a novel joint model that integrates time-to-event and longitudinal TB data, classifying patients into a change-point group or a stable group. For the change-point group, our approach flexibly estimates an individualized change point by leveraging time-to-event information. We use a Monte Carlo Expectation-Maximization (MCEM) algorithm for efficient parameter estimation. Simulation studies demonstrate that our model outperforms traditional approaches by accurately capturing diverse disease progression patterns and handling censoring complexities, leading to robust marginal TB outcome estimates. When applied to a Phase 3 NSCLC trial comparing cemiplimab monotherapy to SOC, the treatment group shows prolonged TB reduction and consistently lower TB over time, highlighting the clinical utility of our approach. The implementation code is publicly available on https://github.com/quyixiang/JoCuR.

stat.ME↗

arXiv2025

Assessing treatment efficacy for interval-censored endpoints using multistate semi-Markov models fit to multiple data streams

We introduce a computationally efficient and general approach for utilizing multiple, possibly interval-censored, data streams to study complex biomedical endpoints using multistate semi-Markov models. Our motivating application is the REGEN-2069 trial, which investigated the protective efficacy (PE) of the monoclonal antibody combination REGEN-COV against SARS-CoV-2 when administered prophylactically to individuals in households at high risk of secondary transmission. Using data on symptom onset, episodic RT-qPCR sampling, and serological testing, we estimate the PE of REGEN-COV for asymptomatic infection, its effect on seroconversion following infection, and the duration of viral shedding. We find that REGEN-COV reduced the risk of asymptomatic infection and the duration of viral shedding, and led to lower rates of seroconversion among asymptomatically infected participants. Our algorithm for fitting semi-Markov models to interval-censored data employs a Monte Carlo expectation maximization (MCEM) algorithm combined with importance sampling to efficiently address the intractability of the marginal likelihood when data are intermittently observed. Our algorithm provide substantial computational improvements over existing methods and allows us to fit semi-parametric models despite complex coarsening of the data.

stat.ME↗

arXiv2025

MLAN: Language-Based Instruction Tuning Preserves and Transfers Knowledge in Multimodal Language Models

We present a novel visual instruction tuning strategy to improve the zero-shot task generalization of multimodal large language models by building a firm text-only knowledge base. Existing work lacks sufficient experimentation on the importance of each modality in the instruction tuning stage, often using a majority of vision-language data while keeping text-only data limited and fixing mixtures of modalities. By incorporating diverse text-only data in the visual instruction tuning stage, we vary vision-language data in various controlled experiments to investigate the importance of modality in visual instruction tuning. Our comprehensive evaluation shows that the text-heavy instruction tuning approach is able to perform on-par with traditional vision-heavy mixtures on both modalities across 12 general datasets while using as low as half the total training tokens. We find that simply increasing sufficiently diverse text-only data enables transfer of instruction following ability and domain knowledge across modalities while being more efficient than the vision-language approach.

cs.CL↗