arXiv subjects

Chenguang Wang

Publications and source records attributed to Chenguang Wang.

At least 73 records · Page 4Linked to original sources

arXiv2025

CaughtCheating: Is Your MLLM a Good Cheating Detective? Exploring the Boundary of Visual Perception and Reasoning

Recent agentic Multi-Modal Large Language Models (MLLMs) such as GPT-o3 have achieved near-ceiling scores on various existing benchmarks, motivating a demand for more challenging test tasks. These MLLMs have been reported to excel in a few expert-level tasks for humans, e.g., GeoGuesser, reflecting their potential as a detective who can notice minuscule cues in an image and weave them into coherent, situational explanations, leading to a reliable answer. But can they match the performance of excellent human detectives? To answer this question, we investigate some hard scenarios where GPT-o3 can still handle, and find a common scenario where o3's performance drops to nearly zero, which we name CaughtCheating. It is inspired by the social media requests that ask others to detect suspicious clues from photos shared by the poster's partner. We conduct extensive experiments and analysis to understand why existing MLLMs lack sufficient capability to solve this kind of task. CaughtCheating provides a class of challenging visual perception and reasoning tasks with great value and practical usage. Success in these tasks paves the way for MLLMs to acquire human-level detective perception and reasoning capabilities.

cs.CV↗

arXiv2025

AgentVigil: Generic Black-Box Red-teaming for Indirect Prompt Injection against LLM Agents

The strong planning and reasoning capabilities of Large Language Models (LLMs) have fostered the development of agent-based systems capable of leveraging external tools and interacting with increasingly complex environments. However, these powerful features also introduce a critical security risk: indirect prompt injection, a sophisticated attack vector that compromises the core of these agents, the LLM, by manipulating contextual information rather than direct user prompts. In this work, we propose a generic black-box fuzzing framework, AgentVigil, designed to automatically discover and exploit indirect prompt injection vulnerabilities across diverse LLM agents. Our approach starts by constructing a high-quality initial seed corpus, then employs a seed selection algorithm based on Monte Carlo Tree Search (MCTS) to iteratively refine inputs, thereby maximizing the likelihood of uncovering agent weaknesses. We evaluate AgentVigil on two public benchmarks, AgentDojo and VWA-adv, where it achieves 71% and 70% success rates against agents based on o3-mini and GPT-4o, respectively, nearly doubling the performance of baseline attacks. Moreover, AgentVigil exhibits strong transferability across unseen tasks and internal LLMs, as well as promising results against defenses. Beyond benchmark evaluations, we apply our attacks in real-world environments, successfully misleading agents to navigate to arbitrary URLs, including malicious sites.

cs.CR↗

arXiv2025

Integrating tumor burden with survival outcome for treatment effect evaluation in oncology trials

In early-phase cancer clinical trials, the limited availability of data presents significant challenges in developing a framework to efficiently quantify treatment effectiveness. To address this, we propose a novel utility-based Bayesian approach for assessing treatment effects in these trials, where data scarcity is a major concern. Our approach synthesizes tumor burden, a key biomarker for evaluating patient response to oncology treatments, and survival outcome, a widely used endpoint for assessing clinical benefits, by jointly modeling longitudinal and survival data. The proposed method, along with its novel estimand, aims to efficiently capture signals of treatment efficacy in early-phase studies and holds potential for development as an endpoint in Phase 3 confirmatory studies. We conduct simulations to investigate the frequentist characteristics of the proposed estimand in a simple setting, which demonstrate relatively controlled Type I error rates when testing the treatment effect on outcomes.

stat.ME↗

arXiv2025

Mosaic-IT: Cost-Free Compositional Data Synthesis for Instruction Tuning

Finetuning large language models with a variety of instruction-response pairs has enhanced their capability to understand and follow instructions. Current instruction tuning primarily relies on teacher models or human intervention to generate and refine the instructions and responses for training, which are costly, non-sustainable, and may lack diversity. In this paper, we introduce Mosaic Instruction Tuning (Mosaic-IT), a human/model-free compositional data synthesis method that can efficiently create rich and diverse augmentations from existing instruction tuning data to enhance the LLMs. Mosaic-IT randomly concatenates multiple instruction data into one and trains the model to produce the corresponding responses with predefined higher-level meta-instructions to strengthen its multi-step instruction-following and format-following skills. Our extensive evaluations demonstrate a superior performance and training efficiency of Mosaic-IT, which achieves consistent performance improvements over various benchmarks and an 80% reduction in training costs compared with original instruction tuning. Our codes and data are available at https://github.com/tianyi-lab/Mosaic-IT.

cs.CL↗

arXiv2025

From Perceptions to Decisions: Wildfire Evacuation Decision Prediction with Behavioral Theory-informed LLMs

Evacuation decision prediction is critical for efficient and effective wildfire response by helping emergency management anticipate traffic congestion and bottlenecks, allocate resources, and minimize negative impacts. Traditional statistical methods for evacuation decision prediction fail to capture the complex and diverse behavioral logic of different individuals. In this work, for the first time, we introduce FLARE, short for facilitating LLM for advanced reasoning on wildfire evacuation decision prediction, a Large Language Model (LLM)-based framework that integrates behavioral theories and models to streamline the Chain-of-Thought (CoT) reasoning and subsequently integrate with memory-based Reinforcement Learning (RL) module to provide accurate evacuation decision prediction and understanding. Our proposed method addresses the limitations of using existing LLMs for evacuation behavioral predictions, such as limited survey data, mismatching with behavioral theory, conflicting individual preferences, implicit and complex mental states, and intractable mental state-behavior mapping. Experiments on three post-wildfire survey datasets show an average of 20.47% performance improvement over traditional theory-informed behavioral models, with strong cross-event generalizability. Our complete code is publicly available at https://github.com/SusuXu-s-Lab/FLARE

cs.AI↗

arXiv2025

COSMIC: Generalized Refusal Direction Identification in LLM Activations

Large Language Models (LLMs) encode behaviors such as refusal within their activation space, yet identifying these behaviors remains a significant challenge. Existing methods often rely on predefined refusal templates detectable in output tokens or require manual analysis. We introduce \textbf{COSMIC} (Cosine Similarity Metrics for Inversion of Concepts), an automated framework for direction selection that identifies viable steering directions and target layers using cosine similarity - entirely independent of model outputs. COSMIC achieves steering performance comparable to prior methods without requiring assumptions about a model's refusal behavior, such as the presence of specific refusal tokens. It reliably identifies refusal directions in adversarial settings and weakly aligned models, and is capable of steering such models toward safer behavior with minimal increase in false refusals, demonstrating robustness across a wide range of alignment conditions.

cs.CL↗

arXiv2025

Importance Weighted Score Matching for Diffusion Samplers with Enhanced Mode Coverage

Training neural samplers directly from unnormalized densities without access to target distribution samples presents a significant challenge. A critical desideratum in these settings is achieving comprehensive mode coverage, ensuring the sampler captures the full diversity of the target distribution. However, prevailing methods often circumvent the lack of target data by optimizing reverse KL-based objectives. Such objectives inherently exhibit mode-seeking behavior, potentially leading to incomplete representation of the underlying distribution. While alternative approaches strive for better mode coverage, they typically rely on implicit mechanisms like heuristics or iterative refinement. In this work, we propose a principled approach for training diffusion-based samplers by directly targeting an objective analogous to the forward KL divergence, which is conceptually known to encourage mode coverage. We introduce \textit{Importance Weighted Score Matching}, a method that optimizes this desired mode-covering objective by re-weighting the score matching loss using tractable importance sampling estimates, thereby overcoming the absence of target distribution data. We also provide theoretical analysis of the bias and variance for our proposed Monte Carlo estimator and the practical loss function used in our method. Experiments on increasingly complex multi-modal distributions, including 2D Gaussian Mixture Models with up to 120 modes and challenging particle systems with inherent symmetries -- demonstrate that our approach consistently outperforms existing neural samplers across all distributional distance metrics, achieving state-of-the-art results on all benchmarks.

cs.LG↗

arXiv2025

Sampling from Binary Quadratic Distributions via Stochastic Localization

Sampling from binary quadratic distributions (BQDs) is a fundamental but challenging problem in discrete optimization and probabilistic inference. Previous work established theoretical guarantees for stochastic localization (SL) in continuous domains, where MCMC methods efficiently estimate the required posterior expectations during SL iterations. However, achieving similar convergence guarantees for discrete MCMC samplers in posterior estimation presents unique theoretical challenges. In this work, we present the first application of SL to general BQDs, proving that after a certain number of iterations, the external field of posterior distributions constructed by SL tends to infinity almost everywhere, hence satisfy Poincaré inequalities with probability near to 1, leading to polynomial-time mixing. This theoretical breakthrough enables efficient sampling from general BQDs, even those that may not originally possess fast mixing properties. Furthermore, our analysis, covering enormous discrete MCMC samplers based on Glauber dynamics and Metropolis-Hastings algorithms, demonstrates the broad applicability of our theoretical framework. Experiments on instances with quadratic unconstrained binary objectives, including maximum independent set, maximum cut, and maximum clique problems, demonstrate consistent improvements in sampling efficiency across different discrete MCMC samplers.

math.ST↗

arXiv2025

Efficient Training of Multi-task Neural Solver for Combinatorial Optimization

Efficiently training a multi-task neural solver for various combinatorial optimization problems (COPs) has been less studied so far. Naive application of conventional multi-task learning approaches often falls short in delivering a high-quality, unified neural solver. This deficiency primarily stems from the significant computational demands and a lack of adequate consideration for the complexities inherent in COPs. In this paper, we propose a general and efficient training paradigm to deliver a unified combinatorial multi-task neural solver. To this end, we resort to the theoretical loss decomposition for multiple tasks under an encoder-decoder framework, which enables more efficient training via proper bandit task-sampling algorithms through an intra-task influence matrix. By employing theoretically grounded approximations, our method significantly enhances overall performance, regardless of whether it is within constrained training budgets, across equivalent training epochs, or in terms of generalization capabilities, when compared to conventional training schedules. On the real-world datasets of TSPLib and CVRPLib, our method also achieved the best results compared to single task learning and multi-task learning approaches. Additionally, the influence matrix provides empirical evidence supporting common practices in the field of learning to optimize, further substantiating the effectiveness of our approach. Our code is open-sourced and available at https://github.com/LOGO-CUHKSZ/MTL-COP.

cs.LG↗

arXiv2025

Where You Go is Who You Are: Behavioral Theory-Guided LLMs for Inverse Reinforcement Learning

Big trajectory data hold great promise for human mobility analysis, but their utility is often constrained by the absence of critical traveler attributes, particularly sociodemographic information. While prior studies have explored predicting such attributes from mobility patterns, they often overlooked underlying cognitive mechanisms and exhibited low predictive accuracy. This study introduces SILIC, short for Sociodemographic Inference with LLM-guided Inverse Reinforcement Learning (IRL) and Cognitive Chain Reasoning (CCR), a theoretically grounded framework that leverages LLMs to infer sociodemographic attributes from observed mobility patterns by capturing latent behavioral intentions and reasoning through psychological constructs. Particularly, our approach explicitly follows the Theory of Planned Behavior (TPB), a foundational behavioral framework in transportation research, to model individuals' latent cognitive processes underlying travel decision-making. The LLMs further provide heuristic guidance to improve IRL reward function initialization and update by addressing its ill-posedness and optimization challenges arising from the vast and unstructured reward space. Evaluated in the 2017 Puget Sound Regional Council Household Travel Survey, our method substantially outperforms state-of-the-art baselines and shows great promise for enriching big trajectory data to support more behaviorally grounded applications in transportation planning and beyond.

cs.AI↗

arXiv2025

Towards Principled Task Grouping for Multi-Task Learning

Multi-task learning (MTL) aims to leverage shared information among tasks to improve learning efficiency and accuracy. However, MTL often struggles to effectively manage positive and negative transfer between tasks, which can hinder performance improvements. Task grouping addresses this challenge by organizing tasks into meaningful clusters, maximizing beneficial transfer while minimizing detrimental interactions. This paper introduces a principled approach to task grouping in MTL, advancing beyond existing methods by addressing key theoretical and practical limitations. Unlike prior studies, our method offers a theoretically grounded approach that does not depend on restrictive assumptions for constructing transfer gains. We also present a flexible mathematical programming formulation that accommodates a wide range of resource constraints, thereby enhancing its versatility. Experimental results across diverse domains, including computer vision datasets, combinatorial optimization benchmarks, and time series tasks, demonstrate the superiority of our method over extensive baselines, thereby validating its effectiveness and general applicability in MTL without sacrificing efficiency.

cs.LG↗

arXiv2025

Beyond the Heatmap: A Rigorous Evaluation of Component Impact in MCTS-Based TSP Solvers

The ``Heatmap + Monte Carlo Tree Search (MCTS)'' paradigm has recently emerged as a prominent framework for solving the Travelling Salesman Problem (TSP). While considerable effort has been devoted to enhancing heatmap sophistication through advanced learning models, this paper rigorously examines whether this emphasis is justified, critically assessing the relative impact of heatmap complexity versus MCTS configuration. Our extensive empirical analysis across diverse TSP scales, distributions, and benchmarks reveals two pivotal insights: 1) The configuration of MCTS strategies significantly influences solution quality, underscoring the importance of meticulous tuning to achieve optimal results and enabling valid comparisons among different heatmap methodologies. 2) A rudimentary, parameter-free heatmap based on the intrinsic $k$-nearest neighbor structure of TSP instances, when coupled with an optimally tuned MCTS, can match or surpass the performance of more sophisticated, learned heatmaps, demonstrating robust generalizability on problem scale and distribution shift. To facilitate rigorous and fair evaluations in future research, we introduce a streamlined pipeline for standardized MCTS hyperparameter tuning. Collectively, these findings challenge the prevalent assumption that heatmap complexity is the primary determinant of performance, advocating instead for a balanced integration and comprehensive evaluation of both learning and search components within this paradigm. Our code is available at: https://github.com/LOGO-CUHKSZ/rethink_mcts_tsp.

cs.LG↗

arXiv2025

JudgeBench: A Benchmark for Evaluating LLM-based Judges

LLM-based judges have emerged as a scalable alternative to human evaluation and are increasingly used to assess, compare, and improve models. However, the reliability of LLM-based judges themselves is rarely scrutinized. As LLMs become more advanced, their responses grow more sophisticated, requiring stronger judges to evaluate them. Existing benchmarks primarily focus on a judge's alignment with human preferences, but often fail to account for more challenging tasks where crowdsourced human preference is a poor indicator of factual and logical correctness. To address this, we propose a novel evaluation framework to objectively evaluate LLM-based judges. Based on this framework, we propose JudgeBench, a benchmark for evaluating LLM-based judges on challenging response pairs spanning knowledge, reasoning, math, and coding. JudgeBench leverages a novel pipeline for converting existing difficult datasets into challenging response pairs with preference labels reflecting objective correctness. Our comprehensive evaluation on a collection of prompted judges, fine-tuned judges, multi-agent judges, and reward models shows that JudgeBench poses a significantly greater challenge than previous benchmarks, with many strong models (e.g., GPT-4o) performing just slightly better than random guessing. Overall, JudgeBench offers a reliable platform for assessing increasingly advanced LLM-based judges. Data and code are available at https://github.com/ScalerLab/JudgeBench.

cs.AI↗

arXiv2025

Understanding Oversmoothing in Diffusion-Based GNNs From the Perspective of Operator Semigroup Theory

This paper presents an analytical study of the oversmoothing issue in diffusion-based Graph Neural Networks (GNNs). Generalizing beyond extant approaches grounded in random walk analysis or particle systems, we approach this problem through operator semigroup theory. This theoretical framework allows us to rigorously prove that oversmoothing is intrinsically linked to the ergodicity of the diffusion operator. Relying on semigroup method, we can quantitatively analyze the dynamic of graph diffusion and give a specific mathematical form of the smoothing feature by ergodicity and invariant measure of operator, which improves previous works only show existence of oversmoothing. This finding further poses a general and mild ergodicity-breaking condition, encompassing the various specific solutions previously offered, thereby presenting a more universal and theoretically grounded approach to relieve oversmoothing in diffusion-based GNNs. Additionally, we offer a probabilistic interpretation of our theory, forging a link with prior works and broadening the theoretical horizon. Our experimental results reveal that this ergodicity-breaking term effectively mitigates oversmoothing measured by Dirichlet energy, and simultaneously enhances performance in node classification tasks.

cs.LG↗

arXiv2025

RuleR: Improving LLM Controllability by Rule-based Data Recycling

Large language models (LLMs) still lack delicate controllability over their responses, which is critical to enhancing their performance and the user experience. However, curating supervised fine-tuning (SFT) datasets to improve LLM controllability usually relies on human experts or proprietary LLMs, which requires additional costs. To bridge this gap, we propose Rule-based Data Recycling (RuleR), a data augmentation method incorporating multiple constraints into the original data samples according to predefined rules, which creates new training tasks to consolidate the controllability of LLMs. Instead of creating new data from scratch, RuleR "recycles" existing data by simply applying rule-based edits to their responses and appending the rule-instructions in their original instructions. Experimental results demonstrate RuleR's effectiveness in improving LLM controllability while maintaining general instruction-following capabilities.

cs.CL↗

arXiv2024

Preference Poisoning Attacks on Reward Model Learning

Learning reward models from pairwise comparisons is a fundamental component in a number of domains, including autonomous control, conversational agents, and recommendation systems, as part of a broad goal of aligning automated decisions with user preferences. These approaches entail collecting preference information from people, with feedback often provided anonymously. Since preferences are subjective, there is no gold standard to compare against; yet, reliance of high-impact systems on preference learning creates a strong motivation for malicious actors to skew data collected in this fashion to their ends. We investigate the nature and extent of this vulnerability by considering an attacker who can flip a small subset of preference comparisons to either promote or demote a target outcome. We propose two classes of algorithmic approaches for these attacks: a gradient-based framework, and several variants of rank-by-distance methods. Next, we evaluate the efficacy of best attacks in both these classes in successfully achieving malicious goals on datasets from three domains: autonomous control, recommendation system, and textual prompt-response preference learning. We find that the best attacks are often highly successful, achieving in the most extreme case 100\% success rate with only 0.3\% of the data poisoned. However, \emph{which} attack is best can vary significantly across domains. In addition, we observe that the simpler and more scalable rank-by-distance approaches are often competitive with, and on occasion significantly outperform, gradient-based methods. Finally, we show that state-of-the-art defenses against other classes of poisoning attacks exhibit limited efficacy in our setting.

cs.LG↗

arXiv2024

Jointly modeling time-to-event and longitudinal data with individual-specific change points: a case study in modeling tumor burden

In oncology clinical trials, tumor burden (TB) stands as a crucial longitudinal biomarker, reflecting the toll a tumor takes on a patient's prognosis. With certain treatments, the disease's natural progression shows the tumor burden initially receding before rising once more. Biologically, the point of change may be different between individuals and must have occurred between the baseline measurement and progression time of the patient, implying a random effects model obeying a bound constraint. However, in practice, patients may drop out of the study due to progression or death, presenting a non-ignorable missing data problem. In this paper, we introduce a novel joint model that combines time-to-event data and longitudinal data, where the latter is parameterized by a random change point augmented by random pre-slope and post-slope dynamics. Importantly, the model is equipped to incorporate covariates across for the longitudinal and survival models, adding significant flexibility. Adopting a Bayesian approach, we propose an efficient Hamiltonian Monte Carlo algorithm for parameter inference. We demonstrate the superiority of our approach compared to a longitudinal-only model via simulations and apply our method to a data set in oncology.

stat.ME↗

arXiv2024

Agent Instructs Large Language Models to be General Zero-Shot Reasoners

We introduce a method to improve the zero-shot reasoning abilities of large language models on general language understanding tasks. Specifically, we build an autonomous agent to instruct the reasoning process of large language models. We show this approach further unleashes the zero-shot reasoning abilities of large language models to more tasks. We study the performance of our method on a wide set of datasets spanning generation, classification, and reasoning. We show that our method generalizes to most tasks and obtains state-of-the-art zero-shot performance on 20 of the 29 datasets that we evaluate. For instance, our method boosts the performance of state-of-the-art large language models by a large margin, including Vicuna-13b (13.3%), Llama-2-70b-chat (23.2%), and GPT-3.5 Turbo (17.0%). Compared to zero-shot chain of thought, our improvement in reasoning is striking, with an average increase of 10.5%. With our method, Llama-2-70b-chat outperforms zero-shot GPT-3.5 Turbo by 10.2%.

cs.CL↗