arXiv subjects

Hengrui Zhang

Publications and source records attributed to Hengrui Zhang.

At least 19 recordsLinked to original sources

arXiv2026

ReasFlow: Assisting Reasoning-Centric Scientific Discovery in Applied Mathematics via a Knowledge-Based Multi-Agent System

Recent advances in Large Language Models have fueled autonomous AI agents capable of tackling complex scientific tasks, yet existing automated research systems remain predominantly focused on empirically driven domains with quantitative benchmarks, leaving theory-driven discovery, particularly in mathematically grounded disciplines requiring rigorous proofs and synthesis of domain knowledge, largely underexplored. Key challenges include the difficulty of verifying theoretical reasoning at scale, insufficient reasoning ability for autonomous frontier exploration, and a scarcity of procedural heuristics in the literature. We introduce ReasFlow, an end-to-end autonomous agent system for reasoning-centric scientific discovery that operationalizes a collaborative paradigm where the human expert acts as Principal Investigator while the agent executes rigorous derivations as a capable graduate student. ReasFlow incorporates (i) a robust internal verification loop that audits logical coherence and corrects fundamental errors prior to human inspection, and (ii) an automated knowledge retrieval and self-improvement mechanism that proactively surfaces both declarative facts and overlooked procedural heuristics, substantially reducing expert intervention. The system unifies literature synthesis, algorithm design, theorem proving, experimentation, and manuscript preparation in a single system. Deployed to autonomously generate five complete research papers with rigorous theoretical and empirical content from minimal prompts, ReasFlow consistently achieves the highest evaluation scores among state-of-the-art open-access baselines under a curated LLM-based review rubric. ReasFlow is publicly accessible via the ReasLab platform, providing a collaborative workspace for AI-assisted theoretical research. Github repo: https://github.com/reaslab/ReasFlow.git.

cs.AI↗

arXiv2026

Wafer-scale Demonstration of High-voltage beta-Ga2O3 MOSFETs with Excellent Uniformity and over 3kV Breakdown Voltages

This study demonstrates a wafer-scale growth of a 2-inch Si-doped $β$-Ga2O3 (100) epitaxial wafer and the realization of uniform, high-voltage lateral $β$-Ga2O3 MOSFET arrays. The 2-inch homoepitaxial $β$-Ga2O3 (100) film grown by MOCVD exhibit excellent crystalline uniformity with an average rocking curve FWHM of ~27.0 arcsec and a low surface roughness less than 1 nm, alongside a uniform net doping concentration on the value of 4.60 $\times$ 1E17 cm-3. The fabricated MOSFETs deliver a threshold voltage of -31.75 V, a drain-current on/off ratio over 1E9, a specific on-resistance of 126.52 mohm$\cdot$cm2 and breakdown voltage exceeding 3 kV. Statistical analysis across the entire wafer presents good device uniformity, with threshold voltages ranging from -28 V to -36 V, output current densities of 60-75 mA/mm, and a breakdown voltage over 3 kV. These results provide the demonstration using the 2-inch $β$-Ga2O3 epitaxial wafer to realize high-voltage $β$-Ga2O3 MOSFETs with wafer-scale performance uniformity for next-generation power device application.

cond-mat.mtrl-sci↗

arXiv2026

BROS: Bias-Corrected Randomized Subspaces for Memory-Efficient Single-Loop Bilevel Optimization

Stochastic bilevel optimization (SBO) has become a standard framework for hyperparameter learning, data reweighting, representation learning, and data-mixture optimization in deep learning. Existing exact single-loop SBO methods and memory-efficient surrogate SBO methods either create severe memory pressure for large lower-level neural networks or lack competitive convergence guarantees under standard assumptions. In this paper, we propose BROS, a memory-efficient single-loop SBO method with the same convergence rate order as exact single-loop SBO methods. BROS performs lower and auxiliary updates in randomized subspaces with a Rademacher bi-probe correction that recovers an unbiased Hessian-action estimator. We prove that BROS preserves the $\mathcal O(\varepsilon^{-2})$ sample complexity of MA-SOBA for finding an $\varepsilon$-stationary point under only standard assumptions. Experiments on hyper-data cleaning, data-mixture learning, hyper-representation learning, and ViT sample reweighting show that BROS reduces peak memory by up to 44.9% while closely matching full-space baseline performance.

cs.LG↗

arXiv2026

SUDA-Muon: Structural Design Principles and Boundaries for Fully Decentralized Muon

Fully decentralized Muon is difficult because its nonlinear matrix-sign operator does not commute with linear gossip averaging. This makes decentralized Muon a structural design problem: in designing the algorithm, one must distinguish modular components from non-modular ones. We propose \sudamuon{}, which realizes this separation through a unified primal--dual communication template called SUDA; within this template, ED/D$^2$, EXTRA, and gradient tracking become modular backbone choices. We prove a topology-separated non-asymptotic convergence guarantee in the nuclear-norm geometry: the dominant term scales as $\mathcal{O}((1+σ/\sqrt{N})K^{-1/4})$ and does not explicitly involve graph quantities, identifying the communication backbone as the modular axis in the structure design. We then establish two complementary non-modular boundaries. Internally, tracking-before-polarization is necessary for this natural no-tracking variant to avoid non-stationary fixed points under heterogeneous objectives. Externally, in the absence of a central server, a fully decentralized method cannot perform the federated average-then-polarize update; we show that this non-modular local-polarize-then-average design is the essential reason why can fail to exhibit linear speedup. Experiments on CIFAR-100 and GPT-2 fine-tuning support the same picture: the unified template makes different communication algorithms directly comparable. In mild near-IID regimes, the resulting variants perform similarly, while in the more difficult long-horizon non-IID CIFAR-100 setting, \sudamuon{} achieves higher accuracy and lower loss than \textsc{DeMuon}.

math.OC↗

arXiv2026

Blinded Multi-Rater Comparative Evaluation of a Large Language Model and Clinician-Authored Responses in CGM-Informed Diabetes Counseling

Continuous glucose monitoring (CGM) is central to diabetes care, but explaining CGM patterns clearly and empathetically remains time-intensive. Evidence for retrieval-grounded large language model (LLM) systems in CGM-informed counseling remains limited. To evaluate whether a retrieval-grounded LLM-based conversational agent (CA) could support patient understanding of CGM data and preparation for routine diabetes consultations. We developed a retrieval-grounded LLM-based CA for CGM interpretation and diabetes counseling support. The system generated plain-language responses while avoiding individualized therapeutic advice. Twelve CGM-informed cases were constructed from publicly available datasets. Between Oct 2025 and Feb 2026, 6 senior UK diabetes clinicians each reviewed 2 assigned cases and answered 24 questions. In a blinded multi-rater evaluation, each CA-generated and clinician-authored response was independently rated by 3 clinicians on 6 quality dimensions. Safety flags and perceived source labels were also recorded. Primary analyses used linear mixed-effects models. A total of 288 unique responses (144 CA and 144 clinician) generated 864 ratings. The CA received higher quality scores than clinician responses (mean 4.37 vs 3.58), with an estimated mean difference of 0.782 points (95% CI 0.692-0.872; P<.001). The largest differences were for empathy (1.062, 95% CI 0.948-1.177) and actionability (0.992, 95% CI 0.877-1.106). Safety flag distributions were similar, with major concerns rare in both groups (3/432, 0.7% each). Retrieval-grounded LLM systems may have value as adjunct tools for CGM review, patient education, and preconsultation preparation. However, these findings do not support autonomous therapeutic decision-making or unsupervised real-world use.

cs.CL↗

arXiv2026

Rethinking Token-Level Policy Optimization for Multimodal Chain-of-Thought

Multimodal Chain-of-Thought (CoT) reasoning requires large vision-language models to construct reasoning trajectories that interleave perceptual grounding with multi-step inference. However, existing Reinforcement Learning with Verifiable Rewards (RLVR) methods typically optimize reasoning at a coarse granularity, treating CoT uniformly without distinguishing their varying degrees of visual grounding. In this work, we conduct a token-level analysis of multimodal reasoning trajectories and show that successful reasoning is characterized by structured token dynamics reflecting both perceptual grounding and exploratory inference. Building upon this analysis, we propose Perception-Exploration Policy Optimization (PEPO), which derives a perception prior from hidden state similarity and integrates it with token entropy through a smooth gating mechanism to produce token-level advantages. PEPO integrates seamlessly with existing RLVR frameworks such as GRPO and DAPO, requiring neither additional supervision nor auxiliary branches. Extensive experiments across diverse multimodal benchmarks demonstrate consistent and robust improvements over strong RL baselines, spanning geometry reasoning, visual grounding, visual puzzle solving, and few-shot classification, while maintaining stable training dynamics. Code: https://github.com/xzxxntxdy/PEPO

cs.CV↗

arXiv2026

Interface Engineered Moiré Graphene Superlattices: Breaking the Auger Carrier Multiplication Limit for Infrared Single-Photon Detection

Hot electrons undergo Auger scattering during their relaxation process has a multiplication effect,which can generate more electrons above the Fermi level, thus improving the efficiency of photoelectric signal conversion.However,the photo-current gain brought by the Auger carrier multiplication is generally limited with a value less than 5,due to the rapid recombination of photo-generated charge-carriers and the inherently low light absorption of two-dimensional materials.Herein,by twisting graphene to an interlayer angle of 10 o ,we report a layer-dependent electronic correlations leading to an efficient carrier multiplication gain of 10 3 .This is primarily offered by the additional localized density-of-states at interface of the bi-layer 10 o ,moire graphene,and the enhanced interlayer coupling of electron waves in a five-layer moire graphene superlattice structure.Therefore,we can harvest the hot electrons during their energy relaxation through a thermalized optical phonon bottleneck effect.It is this effect that promotes the accumulated hot electrons to achieve a maximum Auger scattering rate ~ 10 10 *ps -1 *cm -2 .Furthermore,the ballistic transport of these hot electrons and Schottky barrier from a 90 nm thick silicon-on-insulator (SOI) silicon effectively block the thermal noise,thus leading to a highly sensitive near-infrared detection characteristic.At a low incident light power of ~ 10 -13 W/cm 2 ,the resulting signal-to-noise ratio is more than 100 dB.The strengthened electromagnetic interaction from highly thermalized optical phonon in stacked moire graphene is utilized in this work.The hot electron multiplication suggests the applicability of Van der Waals moire superlattice architecture for harvesting charge carriers,thus paving the pathway to design infrared single-photon avalanche detectors.

physics.app-ph↗

arXiv2026

ScaleDoc: Scaling LLM-based Predicates over Large Document Collections

Predicates are foundational components in data analysis systems. However, modern workloads increasingly involve unstructured documents, which demands semantic understanding, beyond traditional value-based predicates. Given enormous documents and ad-hoc queries, while Large Language Models (LLMs) demonstrate powerful zero-shot capabilities, their high inference cost leads to unacceptable overhead. Therefore, we introduce \textsc{ScaleDoc}, a novel system that addresses this by decoupling predicate execution into an offline representation phase and an optimized online filtering phase. In the offline phase, \textsc{ScaleDoc} leverages a LLM to generate semantic representations for each document. Online, for each query, it trains a lightweight proxy model on these representations to filter the majority of documents, forwarding only the ambiguous cases to the LLM for final decision. Furthermore, \textsc{ScaleDoc} proposes two core innovations to achieve significant efficiency: (1) a contrastive-learning-based framework that trains the proxy model to generate reliable predicating decision scores; (2) an adaptive cascade mechanism that determines the effective filtering policy while meeting specific accuracy targets. Our evaluations across three datasets demonstrate that \textsc{ScaleDoc} achieves over a 2$\times$ end-to-end speedup and reduces expensive LLM invocations by up to 85\%, making large-scale semantic analysis practical and efficient.

cs.DB↗

arXiv2026

On the Convergence of Stochastic Gradient Descent with Perturbed Forward-Backward Passes

We study stochastic gradient descent (SGD) for composite optimization problems with $N$ sequential operators subject to perturbations in both the forward and backward passes. Unlike classical analyses that treat gradient noise as additive and localized, perturbations to intermediate outputs and gradients cascade through the computational graph, compounding geometrically with the number of operators. We present the first comprehensive theoretical analysis of this setting. Specifically, we characterize how forward and backward perturbations propagate and amplify within a single gradient step, derive convergence guarantees for both general non-convex objectives and functions satisfying the Polyak--Łojasiewicz condition, and identify conditions under which perturbations do not deteriorate the asymptotic convergence order. As a byproduct, our analysis furnishes a theoretical explanation for the gradient spiking phenomenon widely observed in deep learning, precisely characterizing the conditions under which training recovers from spikes or diverges. Experiments on logistic regression with convex and non-convex regularization validate our theories, illustrating the predicted spike behavior and the asymmetric sensitivity to forward versus backward perturbations.

math.OC↗

arXiv2026

Towards Universal Video MLLMs with Attribute-Structured and Quality-Verified Instructions

Universal video understanding requires modeling fine-grained visual and audio information over time in diverse real-world scenarios. However, the performance of existing models is primarily constrained by video-instruction data that represents complex audiovisual content as single, incomplete descriptions, lacking fine-grained organization and reliable annotation. To address this, we introduce: (i) ASID-1M, an open-source collection of one million structured, fine-grained audiovisual instruction annotations with single- and multi-attribute supervision; (ii) ASID-Verify, a scalable data curation pipeline for annotation, with automatic verification and refinement that enforces semantic and temporal consistency between descriptions and the corresponding audiovisual content; and (iii) ASID-Captioner, a video understanding model trained via Supervised Fine-Tuning (SFT) on the ASID-1M. Experiments across seven benchmarks covering audiovisual captioning, attribute-wise captioning, caption-based QA, and caption-based temporal grounding show that ASID-Captioner improves fine-grained caption quality while reducing hallucinations and improving instruction following. It achieves state-of-the-art performance among open-source models and is competitive with Gemini-3-Pro.

cs.CV↗

arXiv2025

Multi-Continental Healthcare Modelling Using Blockchain-Enabled Federated Learning

One of the biggest challenges of building artificial intelligence (AI) model in the healthcare area is the data sharing. Since healthcare data is private, sensitive, and heterogeneous, collecting sufficient data for modelling is exhausting, costly, and sometimes impossible. In this paper, we propose a framework for global healthcare modelling using datasets from multi-continents (Europe, North America, and Asia) without sharing the local datasets, and choose glucose management as a study model to verify its effectiveness. Technically, blockchain-enabled federated learning is implemented with adaptation to meet the privacy and safety requirements of healthcare data, meanwhile, it rewards honest participation and penalizes malicious activities using its on-chain incentive mechanism. Experimental results show that the proposed framework is effective, efficient, and privacy-preserving. Its prediction accuracy consistently outperforms models trained on limited personal data and achieves comparable or even slightly better results than centralized training in certain scenarios, all while preserving data privacy. This work paves the way for international collaborations on healthcare projects, where additional data is crucial for reducing bias and providing benefits to humanity.

cs.LG↗

arXiv2025

SPAD: Specialized Prefill and Decode Hardware for Disaggregated LLM Inference

Large Language Models (LLMs) have gained popularity in recent years, driving up the demand for inference. LLM inference is composed of two phases with distinct characteristics: a compute-bound prefill phase followed by a memory-bound decode phase. To efficiently serve LLMs, prior work proposes prefill-decode disaggregation to run each phase on separate hardware. However, existing hardware poorly matches the different requirements of each phase. Current datacenter GPUs and TPUs follow a more-is-better design philosophy that maximizes compute and memory resources, causing memory bandwidth underutilization in the prefill phase and compute underutilization in the decode phase. Such underutilization directly translates into increased serving costs. This paper proposes SPAD (Specialized Prefill and Decode hardware), adopting a less-is-more methodology to design specialized chips tailored to the distinct characteristics of prefill and decode phases. The proposed Prefill Chips have larger systolic arrays and use cost-effective GDDR memory, whereas the proposed Decode Chips retain high memory bandwidth but reduce compute capacity. Compared to modeled H100s, simulations show that the proposed Prefill Chips deliver 8% higher prefill performance on average at 52% lower hardware cost, while the proposed Decode Chips achieve 97% of the decode performance with 28% lower TDP. End-to-end simulations on production traces show that SPAD reduces hardware cost by 19%-41% and TDP by 2%-17% compared to modeled baseline clusters while offering the same performance. Even when models and workloads change, SPAD can reallocate either type of chip to run either phase and still achieve 11%-43% lower hardware costs, demonstrating the longevity of the SPAD design.

cs.AR↗

arXiv2025

A Diffusion-Based Method for Learning the Multi-Outcome Distribution of Medical Treatments

In medicine, treatments often influence multiple, interdependent outcomes, such as primary endpoints, complications, adverse events, or other secondary endpoints. Hence, to make optimal treatment decisions, clinicians are interested in learning the distribution of multi-dimensional treatment outcomes. However, the vast majority of machine learning methods for predicting treatment effects focus on single-outcome settings, despite the fact that medical data often include multiple, interdependent outcomes. To address this limitation, we propose a novel diffusion-based method called DIME to learn the joint distribution of multiple outcomes of medical treatments. We addresses three challenges relevant in medical practice: (i)it is tailored to learn the joint interventional distribution of multiple medical outcomes, which enables reliable decision-making with uncertainty quantification rather than relying solely on point estimates; (ii)it explicitly captures the dependence structure between outcomes; (iii)it can handle outcomes of mixed type, including binary, categorical, and continuous variables. In DIME, we take into account the fundamental problem of causal inference through causal masking. For training, our method decomposes the joint distribution into a series of conditional distributions with a customized conditional masking to account for the dependence structure across outcomes. For inference, our method auto-regressively generates predictions. This allows our method to move beyond point estimates of causal quantities and thus learn the joint interventional distribution. To the best of our knowledge, DIME is the first neural method tailored to learn the joint, multi-outcome distribution of medical treatments. Across various experiments, we demonstrate that our method effectively learns the joint distribution and captures shared information among multiple outcomes.

cs.LG↗

arXiv2025

MUSE: Model-Agnostic Tabular Watermarking via Multi-Sample Selection

We introduce MUSE, a watermarking algorithm for tabular generative models. Previous approaches typically leverage DDIM invertibility to watermark tabular diffusion models, but tabular diffusion models exhibit significantly poorer invertibility compared to other modalities, compromising performance. Simultaneously, tabular diffusion models require substantially less computation than other modalities, enabling a multi-sample selection approach to tabular generative model watermarking. MUSE embeds watermarks by generating multiple candidate samples and selecting one based on a specialized scoring function, without relying on model invertibility. Our theoretical analysis establishes the relationship between watermark detectability, candidate count, and dataset size, allowing precise calibration of watermarking strength. Extensive experiments demonstrate that MUSE achieves state-of-the-art watermark detectability and robustness against various attacks while maintaining data quality, and remains compatible with any tabular generative model supporting repeated sampling, effectively addressing key challenges in tabular data watermarking. Specifically, it reduces the distortion rates on fidelity metrics by 81-89%, while achieving a 1.0 TPR@0.1%FPR detection rate. Implementation of MUSE can be found at https://github.com/fangliancheng/MUSE.

cs.CR↗

arXiv2025

ROAD: Responsibility-Oriented Reward Design for Reinforcement Learning in Autonomous Driving

Reinforcement learning (RL) in autonomous driving employs a trial-and-error mechanism, enhancing robustness in unpredictable environments. However, crafting effective reward functions remains challenging, as conventional approaches rely heavily on manual design and demonstrate limited efficacy in complex scenarios. To address this issue, this study introduces a responsibility-oriented reward function that explicitly incorporates traffic regulations into the RL framework. Specifically, we introduced a Traffic Regulation Knowledge Graph and leveraged Vision-Language Models alongside Retrieval-Augmented Generation techniques to automate reward assignment. This integration guides agents to adhere strictly to traffic laws, thus minimizing rule violations and optimizing decision-making performance in diverse driving conditions. Experimental validations demonstrate that the proposed methodology significantly improves the accuracy of assigning accident responsibilities and effectively reduces the agent's liability in traffic incidents.

cs.LG↗

arXiv2025

DiffPuter: Empowering Diffusion Models for Missing Data Imputation

Generative models play an important role in missing data imputation in that they aim to learn the joint distribution of full data. However, applying advanced deep generative models (such as Diffusion models) to missing data imputation is challenging due to 1) the inherent incompleteness of the training data and 2) the difficulty in performing conditional inference from unconditional generative models. To deal with these challenges, this paper introduces DiffPuter, a tailored diffusion model combined with the Expectation-Maximization (EM) algorithm for missing data imputation. DiffPuter iteratively trains a diffusion model to learn the joint distribution of missing and observed data and performs an accurate conditional sampling to update the missing values using a tailored reversed sampling strategy. Our theoretical analysis shows that DiffPuter's training step corresponds to the maximum likelihood estimation of data density (M-step), and its sampling step represents the Expected A Posteriori estimation of missing values (E-step). Extensive experiments across ten diverse datasets and comparisons with 17 different imputation methods demonstrate DiffPuter's superior performance. Notably, DiffPuter achieves an average improvement of 6.94% in MAE and 4.78% in RMSE compared to the most competitive existing method.

cs.LG↗

arXiv2025

TabGen-ICL: Residual-Aware In-Context Example Selection for Tabular Data Generation

Large Language models (LLMs) have achieved encouraging results in tabular data generation. However, existing approaches require fine-tuning, which is computationally expensive. This paper explores an alternative: prompting a fixed LLM with in-context examples. We observe that using randomly selected in-context examples hampers the LLM's performance, resulting in sub-optimal generation quality. To address this, we propose a novel in-context learning framework: TabGen-ICL, to enhance the in-context learning ability of LLMs for tabular data generation. TabGen-ICL operates iteratively, retrieving a subset of real samples that represent the residual between currently generated samples and true data distributions. This approach serves two purposes: locally, it provides more effective in-context learning examples for the LLM in each iteration; globally, it progressively narrows the gap between generated and real data. Extensive experiments on five real-world tabular datasets demonstrate that TabGen-ICL significantly outperforms the random selection strategy. Specifically, it reduces the error rate by a margin of $3.5\%-42.2\%$ on fidelity metrics. We demonstrate for the first time that prompting a fixed LLM can yield high-quality synthetic tabular data. The code is provided in the \href{https://github.com/fangliancheng/TabGEN-ICL}{link}.

cs.LG↗

arXiv2025

TabDiff: a Mixed-type Diffusion Model for Tabular Data Generation

Synthesizing high-quality tabular data is an important topic in many data science tasks, ranging from dataset augmentation to privacy protection. However, developing expressive generative models for tabular data is challenging due to its inherent heterogeneous data types, complex inter-correlations, and intricate column-wise distributions. In this paper, we introduce TabDiff, a joint diffusion framework that models all mixed-type distributions of tabular data in one model. Our key innovation is the development of a joint continuous-time diffusion process for numerical and categorical data, where we propose feature-wise learnable diffusion processes to counter the high disparity of different feature distributions. TabDiff is parameterized by a transformer handling different input types, and the entire framework can be efficiently optimized in an end-to-end fashion. We further introduce a mixed-type stochastic sampler to automatically correct the accumulated decoding error during sampling, and propose classifier-free guidance for conditional missing column value imputation. Comprehensive experiments on seven datasets demonstrate that TabDiff achieves superior average performance over existing competitive baselines across all eight metrics, with up to $22.5\%$ improvement over the state-of-the-art model on pair-wise column correlation estimations. Code is available at https://github.com/MinkaiXu/TabDiff.

cs.LG↗