arXiv subjects

Dong Xie

Publications and source records attributed to Dong Xie.

At least 19 recordsLinked to original sources

arXiv2026

Sparse Evidence Can Suffice: Agentic Evidence Seeking for Multimodal Video Misinformation Detection

Multimodal video misinformation detection is commonly formulated as a holistic video-understanding task, where the entire video and its associated content are processed and judged in a single pass. However, real-world misinformation often exhibits a sparse and compositional evidence structure: a reliable decision may depend on only a few coupled clues, while most video content contributes limited additional information. Exhaustive multimodal reasoning may therefore introduce substantial redundancy and obscure decisive evidence. This motivates decoupling evidence acquisition from verification: first identifying sparse, decision-relevant clues and then judging veracity based on the acquired evidence. Accordingly, we propose SIEVE, a framework for Sparse Interactive Evidence Verification via Extraction in multimodal video misinformation detection. An evidence-seeking agent actively explores the available multimodal evidence and constructs a compact evidence package, which is then used by a verifier to determine veracity. The agent is trained with supervised evidence-seeking trajectories and an evidence-aware reinforcement learning objective that promotes informative evidence acquisition while discouraging unnecessary or invalid interactions. Experiments on multiple video misinformation benchmarks show that SIEVE consistently outperforms the evaluated baselines and supports reliable verification using compact evidence packages. Moreover, the resulting acquisition process provides an explicit and inspectable evidence trail, improving the transparency and groundedness of multimodal misinformation detection.

cs.CV↗

arXiv2026

Dissipativity-Based Multiport Stability Root-Cause Identification and Mitigation for Solid-State Transformers

For solid-state transformers (SSTs) in high-power grid-connected applications, improperly designed control loops can excite strong inherent AC-DC port coupling, leading to low-frequency oscillation issues, especially under weak grid conditions. To address this problem, this article establishes a multiport admittance matrix for the SST, encompassing its AC dq axes and primary DC port, to characterize its inherent dynamics. Subsequently, a multiport dissipativity analysis is conducted to evaluate the robust stability of the SST. By leveraging the decomposition of passivity conditions into distinct self- and coupling-dissipativity indices, the specific root causes of instability are diagnosed. This framework reveals that a severe coupling-dissipativity failure, induced by the internal dynamics of the synchronization loop, is the dominant instability mechanism rather than a localized self-dissipativity issue. Guided by this diagnosis, a stabilizing controller featuring dynamics-free orthogonal signal reconstruction is designed to reshape the admittance characteristics of the SST. This enhancement specifically targets the identified coupling-dissipativity deficiencies, thereby resolving the root cause of the instability. Finally, the stability analysis and the effectiveness of the enhancement strategy are validated on a down-scaled SST prototype. Experimental results demonstrate that the criterion accurately predicts the coupling-induced oscillations and that the enhanced controller guarantees stable operation under challenging weak-grid conditions.

eess.SY↗

arXiv2026

ReportQA: QA-Based Radiology Report Evaluation

Radiology report evaluation is essential for advancing automated report generation. Natural language generation metrics have limited clinical relevance. Clinical efficacy (CE) metrics evaluate important medical findings, but focus mainly on presence and cover only a limited set of entities. Due to heavy reliance on manual annotations, it is difficult for CE metrics to extend clinical entities or attributes. In clinical practice, radiology reports serve as a medium for information transfer. Clinicians use them to perform downstream diagnostic tasks without directly inspecting images. Based on this insight, we propose ReportQA, a clinical-related and flexible radiology report evaluation framework, supporting detailed quantitative analysis of radiology report generation systems. We first collect datasets covering multiple imaging modalities and anatomical regions. We then construct knowledge trees of clinical entities and attributes with radiologist guidance, and use large language models (LLMs) to extract structured information from raw reports. Next, we generate QA pairs from predefined templates and apply quality control through self-filtering and report-based filtering. During evaluation, the report is treated as context, and an LLM acts as a judge model to answer the QA pairs. Based on the resulting QA accuracy, we introduce QAScore metric. Compared with existing metrics, QAScore shows better alignment with radiologist judgments. Experiments on multiple state-of-the-art vision-language models reveal that current report-based inference paradigms struggle to learn fine-grained clinical representations and exhibit strong negative prior biases. In contrast, question-driven inference provides a more effective alternative. For reproducibility and extensibility, we release the knowledge trees, structured reports, and QA pairs, along with the pipeline code for QA construction and evaluation.

cs.CL↗

arXiv2026

Nonreciprocal Dispersive Coupling for Quantum Sensing

Dispersive coupling is widely utilized for quantum information readout. Most prior studies have concentrated on reciprocal dispersive coupling. Here, we further construct nonreciprocal dispersive coupling and apply it to quantum sensing. For cavity photon number measurement, nonreciprocal dispersive coupling delivers higher precision than its reciprocal counterpart, and this advantage grows more pronounced with an increase in photon number. When directly measuring the single-photon driving strength, however, nonreciprocal dispersive coupling shows no superiority over reciprocal dispersive coupling. By converting the information of driving strength into cavity photon numbers via our proposed strategy, nonreciprocal dispersive coupling again outperforms reciprocal dispersive coupling in precision, with the advantage becoming increasingly significant at larger driving strength. This work presents a novel method to boost quantum sensing and enable the fabrication of ultra-precise quantum sensors.

quant-ph↗

arXiv2026

MapAgent: An Industrial-Grade Agentic Framework for City-scale Lane-level Map Generation

Lane-level maps are critical infrastructure for autonomous driving and lane-level navigation, yet constructing and maintaining standardized lane networks for hundreds of cities remains highly labor-intensive. Recent end-to-end vectorized mapping methods can predict lane geometry and topology directly from sensor data, but they typically treat mapping specifications and traffic regulations as implicit, dataset-dependent supervision. Moreover, in complex scenes (e.g., worn or missing markings and occlusions), correct lane configurations are often under-determined by visual evidence alone, making specification violations a major source of human post-editing. We propose MapAgent, an industrial-grade agentic architecture that augments a vectorization backbone for specification-compliant lane-map production. Rather than merely adding an agent loop to map prediction, MapAgent couples backbone perception with explicit specification verification, constraint-aware reasoning, and deterministic map editing under a bounded, verification-driven Judge-Planner-Worker loop. A vision-language Judge diagnoses errors by jointly inspecting visual evidence and draft vectors, while a tool-calling Planner generates minimal corrective edits with post-edit re-validation. To remain scalable for city-scale production, MapAgent is selectively triggered only on tiles with low backbone confidence, adding modest overhead while preserving throughput. Experiments on real-world datasets show consistent gains over strong production baselines, especially in complex and long-tail scenarios. Additionally, MapAgent has been integrated into Baidu Maps, supporting lane-level map generation for over 360 cities nationwide and elevating the overall production automation to over 95%, demonstrating MapAgent's practicality and effectiveness for large-scale lane-level map generation.

cs.AI↗

arXiv2026

Nonlinear-Gain Distributed Zeroth-Order Optimization for Networked Black-Box Control

This letter studies distributed stochastic optimization over a peer-to-peer network when agents can query only zeroth-order function values. We propose ZOOM-PB, a coordinate-sampling method that blends each local ZO estimate with a fractional-power response while maintaining only a primal state. The raw estimate is retained as a linear anchor, and the nonlinear mixing weight is coupled to the optimization stepsize. This design is motivated by a basic obstruction: transforming heterogeneous or noisy local estimates before averaging can reverse the network direction. We bound that nonlinear residual directly from the raw oracle assumptions instead of imposing an aggregate-alignment condition. With a smooth stochastic-function oracle and a connected graph, ZOOM-PB attains the nonconvex stationarity order $\mathcal{O}(\sqrt{p/(nT)})$ and a Polyak--{\L}ojasiewicz statistical term of order $\mathcal{O}(p/(nT))$, after an explicit initialization transient. Numerical examples compare ZOOM-PB with seven distributed ZO baselines under matched query and message budgets.

eess.SY↗

arXiv2026

Robust Regression with Adaptive Contamination in Response: Optimal Rates and Computational Barriers

We study robust regression under a contamination model in which covariates are clean while the responses may be corrupted in an adaptive manner. Unlike the classical Huber's contamination model, where both covariates and responses may be contaminated and consistent estimation is impossible when the contamination proportion is a non-vanishing constant, it turns out that the clean-covariate setting admits strictly improved statistical guarantees. Specifically, we show that the additional information in the clean covariates can be carefully exploited to construct an estimator that achieves a better estimation rate than that attainable under Huber contamination. In contrast to the Huber model, this improved rate implies consistency even when the contamination is a constant. A matching minimax lower bound is established using Fano's inequality together with the construction of contamination processes that match $m> 2$ distributions simultaneously, extending the previous two-point lower bound argument in Huber's setting. Despite the improvement over the Huber model from an information-theoretic perspective, we provide formal evidence -- in the form of Statistical Query and Low-Degree Polynomial lower bounds -- that the problem exhibits strong information-computation gaps. Our results strongly suggest that the information-theoretic improvements cannot be achieved by polynomial-time algorithms, revealing a fundamental gap between information-theoretic and computational limits in robust regression with clean covariates.

math.ST↗

arXiv2026

Accelerating Fresh Data Exploration with Fluid ETL Pipelines

Recently, we have seen an increasing need for fresh data exploration, where data analysts seek to explore the main characteristics or detect anomalies of data being actively collected. In addition to the common challenges in classic data exploration, such as a lack of prior knowledge about the data or the analysis goal, fresh data exploration also demands an ingestion system with sufficient throughput to keep up with rapid data accumulation. However, leveraging traditional Extract-Transform-Load (ETL) pipelines to achieve low query latency can still be extremely resource-intensive as they must conduct an excessive amount of data preprocessing routines (DPRs) (e.g., parsing and indexing) to cover unpredictable data characteristics and analysis goals. To overcome this challenge, we seek to approach it from a different angle: leveraging occasional idle system capacity or cheap preemptive resources (e.g., Amazon Spot Instance) during ingestion. In particular, we introduce a new type of data ingestion system called fluid ETL pipelines, which allow users to start/stop arbitrary DPRs on demand without blocking data ingestion. With fluid ETL pipelines, users can start potentially useful DPRs to accelerate future exploration queries whenever idle/cheap resources are available. Moreover, users can dynamically change which DPRs to run with limited resources to adapt to users' evolving interests. We conducted experiments on a real-world dataset and verified that our vision is viable. The introduction of fluid ETL pipelines also raises new challenges in handling essential tasks, such as ad-hoc query processing, DPR generation, and DPR management. In this paper, we discuss open research challenges in detail and outline potential directions for addressing them.

cs.DB↗

arXiv2026

On estimating the effective sample size of phylogenetic trees in an autocorrelated chain

Estimating the effective sample size (ESS) is fundamental in Bayesian phylogenetic inference to properly account for autocorrelation in MCMC samples. While methods for continuous parameters are well established, the discrete and high-dimensional nature of treespace poses substantial challenges. Here, we compare existing tree ESS estimators with novel approaches that leverage tractable tree distributions, specifically Conditional Clade Distributions (CCDs), as well as a new probabilistic estimator based on clade frequency differences between independent chains. Using simulated chains with known ESS bounds, we assess estimator accuracy and evaluate their stability and robustness on simulated and real datasets. We further examine how multimodality in posterior distributions and poor mixing can substantially affect ESS estimates, highlighting the need for careful interpretation. Our CCD-based estimators perform comparably to existing approaches, with two methods showing lower variance by averaging across multiple estimates. In contrast, the probabilistic estimator and two previously recommended methods incur prohibitive computational costs for long chains. Together, these results provide guidance for reliable and efficient tree ESS estimation in complex phylogenetic analyses.

q-bio.PE↗

arXiv2026

Beyond End-to-End Video Models: An LLM-Based Multi-Agent System for Educational Video Generation

Although recent end-to-end video generation models demonstrate impressive performance in visually oriented content creation, they remain limited in scenarios that require strict logical rigor and precise knowledge representation, such as instructional and educational media. To address this problem, we propose LASEV, a hierarchical LLM-based multi-agent system for generating high-quality instructional videos from educational problems. LASEV formulates educational video generation as a multi-objective task that simultaneously demands correct step-by-step reasoning, pedagogically coherent narration, semantically faithful visual demonstrations, and precise audio--visual alignment. To address the limitations of prior approaches--including low procedural fidelity, high production cost, and limited controllability--LASEV decomposes the generation workflow into specialized agents that collaborate through a central Orchestrating Agent, shared production state, explicit quality gates, and iterative critique mechanisms. Specifically, the Orchestrating Agent supervises a Solution Agent for rigorous problem solving, an Illustration Agent that produces executable visualization code, and a Narration Agent for learner-oriented instructional scripts. In addition, all outputs from the working agents are subject to semantic critique, rule-based constraints, and tool-based compilation checks. Rather than directly synthesizing pixels, the system constructs a structured executable video script that is deterministically compiled into synchronized visuals and narration using template-driven assembly rules, enabling fully automated production without manual editing. In large-scale deployments, LASEV achieves a throughput exceeding one million videos per day, delivering over a 95% reduction in cost compared to current industry-standard approaches while maintaining a high acceptance rate.

cs.AI↗

arXiv2025

Confidence Intervals for Linear Models with Arbitrary Noise Contamination

We study confidence interval construction for linear regression under Huber's contamination model, where an unknown fraction of noise variables is arbitrarily corrupted. While robust point estimation in this setting is well understood, statistical inference remains challenging, especially because the contamination proportion is not identifiable from the data. We develop a new algorithm that constructs confidence intervals for individual regression coefficients without any prior knowledge of the contamination level. Our method is based on a Z-estimation framework using a smooth estimating function. The method directly quantifies the uncertainty of the estimating equation after a preprocessing step that decorrelates covariates associated with the nuisance parameters. We show that the resulting confidence interval has valid coverage uniformly over all contamination distributions and attains an optimal length of order $O(1/\sqrt{n(1-\epsilon)^2})$, matching the rate achievable when the contamination proportion $\epsilon$ is known. This result stands in sharp contrast to the adaptation cost of robust interval estimation observed in the simpler Gaussian location model.

math.ST↗

arXiv2025

Super-Heisenberg Scaling Using Nonlinear Quantum Scrambling

Super-Heisenberg scaling, which scales as $N^{-\beta}$ with $\beta>1$ in terms of the number of particles $N$ or $T^{-\beta}$ in terms of the evolution time $T$, is better than Heisenberg scaling in quantum metrology. It has been proven that super-Heisenberg scaling can be achieved when the Hamiltonian of the system involves many-body interactions or the time-dependent terms. We demonstrate that nonlinear quantum scrambling facilitates the achievement of super-Heisenberg scaling $T^{-\beta}$ when the generator of the parameter is time-independent. More importantly, in dissipative systems, we can still obtain super-Heisenberg scaling in the friction model. In the optical cavity system, an exponential improvement in measurement precision over time can be achieved by combining injected external squeezing and intracavity squeezing. Our work provides an optimal method for leveraging nonlinear resources to enhance the measurement precision of the driving field.

quant-ph↗

arXiv2025

Photoelastic Grain Solver v2.0: An updated tool for analysis of force measurements in granular materials

Photoelastic force imaging is an experimental technique whereby a birefringent granular material is imaged with a polariscope to characterize the internal stress state of a granular material. Photoelasticimetry is the only proven experimental technique that allows researchers to measure the shear and normal forces at every particle contact in a granular packing. In 2017, Kollmer et al. [Rev. Sci. Instrum. 88, 051808 (2017)] developed an open-source software to perform this analysis. Here, we present the next substantial update to this software package. The new version improves resolution and efficiency and substantially changes the software architecture. The structural changes better facilitate add-ons, modules, and future improvements to the performance, accessibility, and versatility of the tool. Besides updates to the core software, we introduce new infrastructure to support the ongoing development of software, documentation, and training materials. The full development team, software, and supporting resources are available at https://github.com/photoelasticity .

cond-mat.soft↗

arXiv2025

Gravimetry enhanced by nonreciprocal optomechanical coupling

We explore how to measure the gravitational acceleration by using a dissipative optomechanical cavity. What is quite different from the conventional measurement methods is that we have constructed a nonreciprocal optomechanical coupling. We demonstrate that nonreciprocal coupling has a much greater advantage than reciprocal coupling. When the coherent optomechanical coupling is relatively weak and the driving intensity of single photon is strong, nonreciprocal coupling can improve the measurement precision by a factor of two. When the driving strength of single photon tends to infinity, reciprocal coupling fails to obtain any information about the gravitational acceleration, while nonreciprocal coupling still does. Using a two-photon driving, the measurement uncertainty of the gravitational acceleration will tend to zero as the intensity of two-photon driving approaches the critical point. The critical value of the two-photon driving intensity required for nonreciprocal coupling is finite, but the critical value of the two-photon driving intensity required for reciprocal coupling is infinite. The combination of the amplification of mechanical parameters and the extra force can not improve the measurement precision, but it can enhance the susceptibility. Furthermore, when the single-photon driving field is relatively weak, we analytically calculate the quantum Fisher information. The results show that, under most experimental parameters, nonreciprocal coupling still performs better than reciprocal coupling.

quant-ph↗

arXiv2025

Security Analysis of Thumbnail-Preserving Image Encryption and a New Framework

As a primary encryption primitive balancing the privacy and searchability of cloud storage images, thumbnail preserving encryption (TPE) enables users to quickly identify the privacy personal image on the cloud and request this image from the owner through a secure channel. In this paper, we have found that two different plaintext images may produce the same thumbnail. It results in the failure of search strategy because the collision of thumbnail occurs. To address this serious security issues, we conduct an in-depth analysis on the collision probabilities of thumbnails, and then propose a new TPE framework, called multi-factor thumbnail preserving encryption (MFTPE). It starts from the collision probability of two blocks, extend to the probabilities of two images and ultimately to N images. Then, we in detail describe three specific MFTPE constructions preserving different combinations of factors, i.e., the sum and the geometric mean, the sum and the range, and the sum and the weighted mean. The theoretical and experimental results demonstrate that the proposed MFTPE reduces the probability of thumbnails, exhibits strong robustness, and also effectively resists face detection and noise attacks.

cs.CR↗

arXiv2025

Intelligent Soft Matter: Towards Embodied Intelligence

Intelligent soft matter stands at the intersection of materials science, physics, and cognitive science, promising to change how we design and interact with materials. This transformative field seeks to create materials that possess life-like capabilities, such as perception, learning, memory, and adaptive behavior. Unlike traditional materials, which typically perform static or predefined functions, intelligent soft matter dynamically interacts with its environment. It integrates multiple sensory inputs, retains experiences, and makes decisions to optimize its responses. Inspired by biological systems, these materials intend to leverage the inherent properties of soft matter: flexibility, self-evolving, and responsiveness to perform functions that mimic cognitive processes. By synthesizing current research trends and projecting their evolution, we present a forward-looking perspective on how intelligent soft matter could be constructed, with the aim of inspiring innovations in fields such as biomedical devices, adaptive robotics, and beyond. We highlight new pathways for integrating design of sensing, memory and action with internal low-power operations and discuss challenges for practical implementation of materials with "intelligent behavior". These approaches outline a path towards to more robust, versatile and scalable materials that can potentially act, compute, and "think" by their inherent intrinsic material behaviour beyond traditional smart technologies relying on external control.

cond-mat.soft↗

arXiv2024

Dispersive Qubit Readout of Temperature

Squeezed light can exponentially increase the signal-to-noise ratio (SNR) of dispersive qubit readout, especially using a combination of injected external squeezing (IES) and intracavity squeezing (ICS). We further investigate whether IES and ICS can also exponentially improve the measurement precision of temperature. In the case of fully thermalized qubits isolated from thermal bath, the measurement precision of temperature can be improved exponentially when the temperature or measurement time or the input photon number approaches 0. In general, thermal fluctuations prevent the action of squeezed light. When multiple qubits maintain interacting with the thermal bath, the Heisenberg scaling can be achieved if the loss rate of the cavity is large and the coupling between the qubit and the optical cavity is weak enough. In the meantime, IES can also further promote the improvement of the measurement precision of the temperature exponentially.

quant-ph↗

arXiv2024

ARB-LLM: Alternating Refined Binarizations for Large Language Models

Large Language Models (LLMs) have greatly pushed forward advancements in natural language processing, yet their high memory and computational demands hinder practical deployment. Binarization, as an effective compression technique, can shrink model weights to just 1 bit, significantly reducing the high demands on computation and memory. However, current binarization methods struggle to narrow the distribution gap between binarized and full-precision weights, while also overlooking the column deviation in LLM weight distribution. To tackle these issues, we propose ARB-LLM, a novel 1-bit post-training quantization (PTQ) technique tailored for LLMs. To narrow the distribution shift between binarized and full-precision weights, we first design an alternating refined binarization (ARB) algorithm to progressively update the binarization parameters, which significantly reduces the quantization error. Moreover, considering the pivot role of calibration data and the column deviation in LLM weights, we further extend ARB to ARB-X and ARB-RC. In addition, we refine the weight partition strategy with column-group bitmap (CGB), which further enhance performance. Equipping ARB-X and ARB-RC with CGB, we obtain ARB-LLM$_\text{X}$ and ARB-LLM$_\text{RC}$ respectively, which significantly outperform state-of-the-art (SOTA) binarization methods for LLMs. As a binary PTQ method, our ARB-LLM$_\text{RC}$ is the first to surpass FP16 models of the same size. The code and models will be available at https://github.com/ZHITENGLI/ARB-LLM.

cs.CV↗