arXiv subjects

Mingxuan Zhang

Publications and source records attributed to Mingxuan Zhang.

At least 19 recordsLinked to original sources

arXiv2026

RAFT: A Stateful Retrieval-Augmented Framework for Troubleshooting Agents

Effective troubleshooting agents in enterprise customer support depend on retrieving actionable guidance from similar historical cases, yet existing retrieval-augmented generation (RAG) systems treat support cases as static documents and overlook their multi-stage, stateful nature. We introduce RAFT (Retrieval-Augmented Framework for Troubleshooting Agents), a stateful RAG framework that abstracts each closed historical case into a directed chain of timeline entries and retrieves at the entry level, surfacing cases whose intermediate states match the active case and returning the parent-case trajectory anchored at the matched state; an optional case-level graph links cases through a configurable similarity representation. We evaluate this retrieval layer directly, which, unlike evaluating a full agent system, requires no production deployment. Because public multi-stage troubleshooting data is extremely rare, we pair a synthetic benchmark built from Microsoft Learn Windows Server documentation with real Apache Jira issues carrying human-created duplicate labels. RAFT improves Case Hit over vanilla RAG and GraphRAG baselines at every stage of case progress, with statistically significant gains over the strongest baseline; the Jira results provide directional evidence that the advantage transfers to real case histories. We release our benchmark, implementation, and the Apache Jira evaluation set.

cs.AI↗

arXiv2026

Hidden in Plain Sight: Diffusion-Based Unrestricted Robotic Attacks on Vision-Language-Action Models

Vision-Language-Action (VLA) models have shown strong capabilities in controlling robots across diverse manipulation tasks. However, their adversarial robustness remains largely underexplored, and exploiting this weakness can lead to physical-world harm. Existing attacks on VLA models often rely on pixel-space perturbations or white-box access, resulting in noticeable artifacts and limited deployability in real-world robotic systems. In this work, we propose DURA, a diffusion-based unrestricted robotic attack that generates visually natural adversarial patches for VLA models. DURA supports both white-box and black-box attack settings, where the black-box setting requires only the predicted actions of the victim model. By optimizing along the latent trajectory of a pretrained diffusion model, DURA generates visually natural patches while steering the robot toward attacker-specified target actions. Extensive experiments in both simulation and the real physical world show that DURA consistently outperforms existing methods. Our findings expose a safety risk for physically deployed VLA models and call for stronger defenses.

cs.AI↗

arXiv2026

MoE-Prism: Disentangling Monolithic Experts for Elastic MoE Services via Model-System Co-Designs

Mixture-of-Experts (MoE) scales model capacity through sparse activation, and is becoming an important architecture for large language models (LLMs). However, existing MoE serving systems typically execute all requests under a fixed routing configuration, limiting their ability to exploit heterogeneous computation requirements across requests. Routing top-$k$, which determines the number of routed experts activated per token, directly controls routed-expert computation and provides a natural mechanism for request-level compute elasticity. Realizing this capability, however, requires finer-grained routing units and efficient runtime execution for heterogeneous routing budgets. We present \textsc{MoE-Prism}, a model and system support framework for request-level compute elasticity in MoE serving. \textsc{MoE-Prism}decomposes monolithic experts into fine-grained sub-experts to expose denser routing operating points and provides a $k$-aware serving runtime that effectively serves heterogeneous routing budgets under both throughput-oriented and latency-sensitive workloads. We implement \textsc{MoE-Prism} on top of vLLM and evaluate it on three representative MoE models. \textsc{MoE-Prism} expands the number of available routing operating points by $4\times$, improves offline inference throughput by up to 33.9\%, and reduces online serving TTFT under heterogeneous workloads. These results demonstrate practical elastic MoE serving with request-level routing targets.

cs.CL↗

arXiv2026

PrivacyPeek: Auditing What LLM-Based Agents Acquire, Not Just What They Say

LLM-based agents are rapidly advancing, autonomously invoking external tools to complete multi-step tasks for users. However, agents often acquire more sensitive information than the task requires. Existing privacy benchmarks audit what the agent's response or outgoing actions disclose, but overlook the acquisition stage where data first enters the agent's context. The over-acquired information is then one careless action or one attack away from an outright leak. To assess its prevalence, we introduce \emph{PrivacyPeek}, a benchmark for evaluating acquisition-stage privacy leakage of LLM-based agents, with $1{,}182$ cases across $7$ acquisition behaviours and $16$ application domains. Specifically, \emph{Acquisition Inspection} examines the agent's tool-call trajectory, both the tools it invokes and the data it receives, to detect when it acquires sensitive information beyond the task scope. \emph{Probe Elicitation} then issues a follow-up probe and measures how readily an attacker could elicit sensitive information the agent acquired but did not disclose. Our experiments on $10$ LLM-based agents across $4$ model families show that the unnecessary acquisition of sensitive information is widespread. In addition, we observe a correlation between the task-completion capability and acquisition-stage leakage. Prompt-level defences reduce only a small fraction of acquisition-stage leakage, leaving the majority unmitigated. These results make auditing acquisition-stage privacy both urgent and necessary. Our dataset and code are available at https://github.com/Xuan269/PrivacyPeek-Resource.

cs.CR↗

arXiv2026

A fault-tolerant quantum blockchain deployed on commercial telecommunications network

Popularized by the Bitcoin cryptocurrency, blockchain technology establishes a decentralized digital framework that utilizes cryptographic and consensus protocols to secure data against unauthorized modification. Consequently, blockchain has found broad adoption across diverse fields, including finance, data management, healthcare, and digital asset governance. In the quantum computing era, a paramount objective for blockchain is to preserve its foundational advantages of cryptographic integrity and decentralized fault-tolerant resilience. In principle, quantum digital signatures and quantum Byzantine agreement protocols offer foundational security guarantees and tolerate up to one-half of malicious nodes for blockchain. However, the practical realization of such a quantum-enhanced blockchain remains a significant and multifaceted challenge. Here, we propose and experimentally demonstrate a fully operational hybrid quantum blockchain architecture built on photonic integrated circuits and deployed over commercially available classical telecommunications infrastructure. The system achieves a fault tolerance of nearly one-half, surpassing the classical limit, while reaching consensus on a timescale of seconds. A deployed food traceability application validates the practicality of the proposed architecture, achieving a throughput of approximately 500 transactions per second. This work establishes a foundation for practical quantum blockchains, enabling secure, scalable, and decentralized information processing in the emerging quantum era.

quant-ph↗

arXiv2026

Experimental demonstration of scalable quantum blockchain with exponentially superior quantum communication complexity

To secure modern distributed digital infrastructures, quantum blockchains exploit quantum resources to achieve information-theoretic security and surpass the classical one-third fault-tolerance bound. However, existing high-fault-tolerant protocols face a fundamental scalability challenge: the blockchain trilemma imposes either exponential communication complexity or experimentally demanding multipartite entanglement. Here, we experimentally demonstrate a scalable quantum blockchain protocol based on weak coherent states that achieves an exponential reduction in quantum communication complexity. The protocol employs a circular quantum Byzantine agreement mechanism that preserves information-theoretic security while avoiding multipartite entanglement. We implement this protocol on a photonic integrated circuit platform, realizing a six-node network over commercially available telecommunication infrastructure. Compared with previous schemes, the protocol requires less than 4% of the quantum communication resources. Leveraging this advantage, we further demonstrate a quantum-secured token exchange application achieving a throughput of 805.3 transactions per second with zero failures. These results establish a practical pathway toward scalable quantum blockchain.

quant-ph↗

arXiv2026

Learning Lineage-guided Geodesics with Finsler Geometry

Trajectory inference investigates how to interpolate paths between observed timepoints of dynamical systems, such as temporally resolved population distributions, with the goal of inferring trajectories at unseen times and better understanding system dynamics. Previous work has focused on continuous geometric priors, utilizing data-dependent spatial features to define a Riemannian metric. In many applications, there exists discrete, directed prior knowledge over admissible transitions (e.g. lineage trees in developmental biology). We introduce a Finsler metric that combines geometry with classification and incorporate both types of priors in trajectory inference, yielding improved performance on interpolation tasks in synthetic and real-world data.

cs.LG↗

arXiv2026

EVLA: An Electro-Aware Multimodal Assistant for Physically-Grounded Driving Reasoning and Control

Modern vision-language models (VLMs) for driving assistants typically treat vehicle dynamics as a black box, resulting in decisions that lack awareness of the vehicle's real-time electro-mechanical state. To bridge this gap, we introduce the Electro-Visual-Language Assistant (EVLA) -- a novel framework that combines multi-modal scene understanding with real-time perception of the electrified powertrain state (e.g., motor torque, battery SOC). Our approach features two key innovations: first, a Unified Co-State Encoder (UCSE) that fuses visual, textual, and vehicle-state inputs into a shared latent representation, augmented with an Energy-Efficiency Field to model spatial energy costs; and second, an Electro-aware Structured Reasoning Chain (ESRC), which replaces external chain-of-thought prompting with an internal, deterministic reasoning process grounded in physical constraints and optimization objectives. Trained end-to-end with a physics-guided joint loss, EVLA learns to generate context-aware and energy-optimal driving decisions. Extensive evaluations on a driving QA benchmark demonstrate that EVLA substantially outperforms strong fine-tuned VLM baselines, improving the final score by +0.0871 and accuracy by +5.6\%. Ablation studies validate the necessity of each component, and efficiency analyses show that EVLA achieves 36\% faster inference than multi-stage pipelines. This work underscores that integrating vehicle-state awareness and structured physical reasoning is crucial for developing next-generation, physically-grounded driving assistants.

cs.CL↗

arXiv2026

FLAP: FOV-Constrained Active Perception Planning for Prior-Map-Free 3D Navigation

Safe and efficient trajectory planning in unknown, cluttered 3D environments constitutes a critical bottleneck for deploying Unmanned Aerial Vehicles (UAVs) in real-world applications. This challenge is further exacerbated by the limited field-of-view (FOV) and sensing range of onboard sensors. Many existing methods either make simplistic assumptions about unexplored space or rely on conservative heuristics such as speed limits or fixed perception patterns, reducing efficiency and generalizing poorly across different sensor types. In this work, we propose a novel planning framework that directly integrates active perception into trajectory optimization, thereby improving safety while preserving efficiency. The perception constraints are derived from the UAV's dynamic model and formulated in the sensor coordinate frame, which enables precise handling of FOV geometry. The velocity-triggered activation mechanism enables the planner to balance perception and motion efficiency. We introduce an active perception sub-trajectory segment with parametric start-time optimization, mitigating collision risks from late obstacle detection. Our formulation enables active perception during arbitrary 3D maneuvers, extending beyond prior methods designed mainly for horizontal motion. All constraints and penalties are incorporated into a differentiable optimization problem, so the planner requires only a simple front-end global path for guidance, rather than a computationally expensive perception-aware path generator. Extensive simulations and real-world experiments demonstrate robust performance across diverse unknown environments with varying sensor configurations.

cs.RO↗

arXiv2026

High-rate quantum digital signatures over 250 km of optical fiber

Quantum digital signatures (QDS) offer information-theoretic security for message integrity, authenticity, and non-repudiation, and constitute a fundamental cryptographic primitive for future quantum networks. Despite significant progress, the practical deployment of QDS has been severely constrained by limited signature rates and poor tolerance to channel loss, particularly in long-distance and metropolitan-scale networks. Here, we report a high-rate, loss-resilient QDS system that overcomes these two key bottlenecks simultaneously. Our implementation combines intrinsically phase-stable polarization modulation based on a Sagnac interferometer with gigahertz-rate quantum state encoding and low-timing-jitter superconducting nanowire single-photon detectors, enabling robust and continuous operation at high repetition frequencies. By integrating this hardware platform with a one-time universal hashing-based QDS protocol, we achieve a signature rate improvement of more than two orders of magnitude compared with existing QDS implementations under comparable channel-loss conditions. Notably, the system maintains a non-zero effective signature rate of approximately 1.25 times per second at a total channel loss of up to 49.05 dB, representing the highest loss tolerance reported for QDS to date. These results establish a practical and scalable technological pathway for deploying QDS in real-world quantum communication networks.

quant-ph↗

arXiv2026

Scalable Contrastive Causal Discovery under Unknown Soft Interventions

Observational causal discovery is only identifiable up to the Markov equivalence class. While interventions can reduce this ambiguity, in practice interventions are often soft with multiple unknown targets. In many realistic scenarios, only a single intervention regime is observed. We propose a scalable causal discovery model for paired observational and interventional settings with shared underlying causal structure and unknown soft interventions. The model aggregates subset-level PDAGs and applies contrastive cross-regime orientation rules to construct a globally consistent maximal PDAG under Meek closure, enabling generalization to both in-distribution and out-of-distribution settings. Theoretically, we prove that our model is sound with respect to a restricted $Ψ$ equivalence class induced solely by the information available in the subset-restricted setting. We further show that the model asymptotically recovers the corresponding identifiable PDAG and can orient additional edges compared to non-contrastive subset-restricted methods. Experiments on synthetic data demonstrate improved causal structure recovery, generalization to unseen graphs with held-out causal mechanisms, and scalability to larger graphs, with ablations supporting the theoretical results.

stat.ML↗

arXiv2026

Dense Feature Learning via Linear Structure Preservation in Medical Data

Deep learning models for medical data are typically trained using task specific objectives that encourage representations to collapse onto a small number of discriminative directions. While effective for individual prediction problems, this paradigm underutilizes the rich structure of clinical data and limits the transferability, stability, and interpretability of learned features. In this work, we propose dense feature learning, a representation centric framework that explicitly shapes the linear structure of medical embeddings. Our approach operates directly on embedding matrices, encouraging spectral balance, subspace consistency, and feature orthogonality through objectives defined entirely in terms of linear algebraic properties. Without relying on labels or generative reconstruction, dense feature learning produces representations with higher effective rank, improved conditioning, and greater stability across time. Empirical evaluations across longitudinal EHR data, clinical text, and multimodal patient representations demonstrate consistent improvements in downstream linear performance, robustness, and subspace alignment compared to supervised and self supervised baselines. These results suggest that learning to span clinical variation may be as important as learning to predict clinical outcomes, and position representation geometry as a first class objective in medical AI.

cs.LG↗

arXiv2026

PEAR: Planner-Executor Agent Robustness Benchmark

Large Language Model (LLM)-based Multi-Agent Systems (MAS) have emerged as a powerful paradigm for tackling complex, multi-step tasks across diverse domains. However, despite their impressive capabilities, MAS remain susceptible to adversarial manipulation. Existing studies typically examine isolated attack surfaces or specific scenarios, leaving a lack of holistic understanding of MAS vulnerabilities. To bridge this gap, we introduce PEAR, a benchmark for systematically evaluating both the utility and vulnerability of planner-executor MAS. While compatible with various MAS architectures, our benchmark focuses on the planner-executor structure, which is a practical and widely adopted design. Through extensive experiments, we find that (1) a weak planner degrades overall clean task performance more severely than a weak executor; (2) while a memory module is essential for the planner, having a memory module for the executor does not impact the clean task performance; (3) there exists a trade-off between task performance and robustness; and (4) attacks targeting the planner are particularly effective at misleading the system. These findings offer actionable insights for enhancing the robustness of MAS and lay the groundwork for principled defenses in multi-agent settings.

cs.LG↗

arXiv2025

TooBadRL: Trigger Optimization to Boost Effectiveness of Backdoor Attacks on Deep Reinforcement Learning

Deep reinforcement learning (DRL) has achieved remarkable success in a wide range of sequential decision-making applications, including robotics, healthcare, smart grids, and finance. Recent studies reveal that adversaries can implant backdoors into DRL agents during the training phase. These backdoors can later be activated by specific triggers during deployment, compelling the agent to execute targeted actions and potentially leading to severe consequences, such as drone crashes or vehicle collisions. However, existing backdoor attacks utilize simplistic and heuristic trigger configurations, overlooking the critical impact of trigger design on attack effectiveness. To address this gap, we introduce TooBadRL, the first framework to systematically optimize DRL backdoor triggers across three critical aspects: injection timing, trigger dimension, and manipulation magnitude. Specifically, we first introduce a performance-aware adaptive freezing mechanism to determine the injection timing during training. Then, we formulate trigger selection as an influence attribution problem and apply Shapley value analysis to identify the most influential trigger dimension for injection. Furthermore, we propose an adversarial input synthesis method to optimize the manipulation magnitude under environmental constraints. Extensive evaluations on three DRL algorithms and nine benchmark tasks demonstrate that TooBadRL outperforms five baseline methods in terms of attack success rate while only slightly affecting normal task performance. We further evaluate potential defense strategies from detection and mitigation perspectives. We open-source our code to facilitate reproducibility and further research.

cs.CR↗

arXiv2025

MoE-SpeQ: Speculative Quantized Decoding with Proactive Expert Prefetching and Offloading for Mixture-of-Experts

The immense memory requirements of state-of-the-art Mixture-of-Experts (MoE) models present a significant challenge for inference, often exceeding the capacity of a single accelerator. While offloading experts to host memory is a common solution, it introduces a severe I/O bottleneck over the PCIe bus, as the data-dependent nature of expert selection places these synchronous transfers directly on the critical path of execution, crippling performance. This paper argues that the I/O bottleneck can be overcome by trading a small amount of cheap, on-device computation to hide the immense cost of data movement. We present MoE-SpeQ, a new inference system built on a novel co-design of speculative execution and expert offloading. MoE-SpeQ employs a small, on-device draft model to predict the sequence of required experts for future tokens. This foresight enables a runtime orchestrator to prefetch these experts from host memory, effectively overlapping the expensive I/O with useful computation and hiding the latency from the critical path. To maximize performance, an adaptive governor, guided by an Amortization Roofline Model, dynamically tunes the speculation strategy to the underlying hardware. Our evaluation on memory-constrained devices shows that for the Phi-MoE model, MoE-SpeQ achieves at most 2.34x speedup over the state-of-the-art offloading framework. Our work establishes a new, principled approach for managing data-dependent memory access in resource-limited environments, making MoE inference more accessible on commodity hardware.

cs.LG↗

arXiv2025

Energy Guided Geometric Flow Matching

A useful inductive bias for temporal data is that trajectories should stay close to the data manifold. Traditional flow matching relies on straight conditional paths, and flow matching methods which learn geodesics rely on RBF kernels or nearest neighbor graphs that suffer from the curse of dimensionality. We propose to use score matching and annealed energy distillation to learn a metric tensor that faithfully captures the underlying data geometry and informs more accurate flows. We demonstrate the efficacy of this strategy on synthetic manifolds with analytic geodesics, and interpolation of cell

cs.LG↗

arXiv2024

Magnitude Pruning of Large Pretrained Transformer Models with a Mixture Gaussian Prior

Large pretrained transformer models have revolutionized modern AI applications with their state-of-the-art performance in natural language processing (NLP). However, their substantial parameter count poses challenges for real-world deployment. To address this, researchers often reduce model size by pruning parameters based on their magnitude or sensitivity. Previous research has demonstrated the limitations of magnitude pruning, especially in the context of transfer learning for modern NLP tasks. In this paper, we introduce a new magnitude-based pruning algorithm called mixture Gaussian prior pruning (MGPP), which employs a mixture Gaussian prior for regularization. MGPP prunes non-expressive weights under the guidance of the mixture Gaussian prior, aiming to retain the model's expressive capability. Extensive evaluations across various NLP tasks, including natural language understanding, question answering, and natural language generation, demonstrate the superiority of MGPP over existing pruning methods, particularly in high sparsity settings. Additionally, we provide a theoretical justification for the consistency of the sparse transformer, shedding light on the effectiveness of the proposed pruning method.

stat.ML↗

arXiv2024

One-dimensional Photonic Crystal Structure Enhanced External-Magnetic-Field-Free Spintronic Terahertz High-Field Emitter

Intense terahertz (THz) radiation in free space offers multifaceted capabilities for accelerating electron, understanding the mesoscale architecture in (bio)materials, elementary excitation and so on. Recently popularized spintronic THz emitters (STEs) with their versatility such as ultra-broadband, cost-effectiveness, large-size and ease for-integration have become one of the most promising alternative for the next generation of intense THz sources. Nevertheless, the typical W| Co20Fe60B20 | Pt necessitates an external-magnetic-field to saturate magnetization for stable operation, limiting its scalability for achieving higher THz field with uniform distribution over larger sample areas. Here we demonstrate the methodologies of enhancing the high-field THz radiation of external-magnetic-field-free IrMn3 | Co20Fe60B20 |W heterostructure via optimizing the substrate with superior thermal conductivity and integrating a one-dimensional photonic crystal (PC) structure to maximize the radiation efficiency. Under the excitation of a Ti: sapphire femtosecond laser amplifier with central wavelength of 800 nm, pulse duration of 35 fs, and repetition rate of 1 kHz and maximum single pulse energy of 5.5 mJ, we successfully generate intense THz radiation with focal peak electric field up to 1.1 MV/cm with frequency range covering 0.1-10 THz without external-magnetic-fields. These high-field STEs will also enable other applications such as ultra-broadband high-field THz spectroscopy and polarization-based large-size strong-field THz imaging.

physics.optics↗