Searcharxiv⌕ Search

arXiv subjects

Wei Sun

Publications and source records attributed to Wei Sun.

At least 73 records · Page 4Linked to original sources
arXiv2026

Decision-Centric Design for LLM Systems

LLM systems must make control decisions in addition to generating outputs: whether to answer, clarify, retrieve, call tools, repair, or escalate. In many current architectures, these decisions remain implicit within generation, entangling assessment and action in a single model call and making failures hard to inspect, constrain, or repair. We propose a decision-centric framework that separates decision-relevant signals from the policy that maps them to actions, turning control into an explicit and inspectable layer of the system. This separation supports attribution of failures to signal estimation, decision policy, or execution, and enables modular improvement of each component. It unifies familiar single-step settings such as routing and adaptive inference, and extends naturally to sequential settings in which actions alter the information available before acting. Across three controlled experiments, the framework reduces futile actions, improves task success, and reveals interpretable failure modes. More broadly, it offers a general architectural principle for building more reliable, controllable, and diagnosable LLM systems.

cs.AI↗
arXiv2026

1-Mbps Twin-Field Quantum Key Distribution over 200 km Using Independent Dissipative Kerr Solitons

Twin-field quantum key distribution (TF-QKD) dramatically enhances the secure key rate (SKR) over inter-city distances through its square-root scaling. Further improvements in aggregate SKR can be achieved by wavelength-division multiplexing (WDM) of parallel QKD channels. However, direct implementation in TF-QKD poses significant challenges, as each wavelength channel requires an independent ultra-stable seed laser, narrow-linewidth transmitters, and optical phase-locked loops (OPLLs), which are not easily scalable. Here, we circumvent these limitations by employing two independent, integrated dissipative Kerr soliton (DKS) microcombs at Alice and Bob as multi-wavelength sources. High-visibility single-photon interference across all wavelength channels is achieved by stabilizing the frequencies of every comb line - requiring only the stabilization of the pump wavelength and repetition rates of the two microcombs. Based on this architecture, we perform a full TF-QKD experiment using the sending-or-not-sending protocol, achieving a total SKR of 1.57 Mbps over 201.1 km of fiber using 16 DWDM channels. This result represents more than an order-of-magnitude enhancement compared with single-wavelength TF-QKD at the same distance. Given that a single DKS comb can support over 100 coherent lines across the C-band, this approach offers a scalable pathway toward high-rate quantum key distribution over inter-city distances.

quant-ph↗
arXiv2026

Kramers-Kronig causality in integrated photonics: The spectral tension between ultraviolet transition and mid-infrared absorption

Dispersion engineering via geometric confinement is essential to integrated photonics, enabling phenomena such as soliton microcombs, supercontinua, parametric oscillators, and entangled photons. However, prevailing methodologies rely on semi-empirical Sellmeier models that assume idealized material purity, neglecting the pronounced dispersion shifts induced by residual impurities like hydrogen-related bonds. Here, we demonstrate that these residual bonds fundamentally alter the dispersion landscape spanning from the ultraviolet (UV) to the mid-infrared (MIR) spectra. Specifically, they introduce MIR vibrational absorption while simultaneously modifying UV electronic transition, shifting the bandgap and UV pole. We show that the spectral tension between these UV and MIR modifications dictates the group velocity dispersion from the visible to the near-infrared (NIR) via the Kramers-Kronig causality. We experimentally validate this phenomenon through systematic characterization of broadband loss and dispersion in ultralow-loss silicon nitride photonic integrated circuits. By rigorously incorporating these effects, we bridge the gap between empirical fitting and predictive physical modelling. Our study resolves long-standing discrepancies in dispersion engineering, providing precision control essential for next-generation integrated photonics.

physics.optics↗
arXiv2026

DP^2-VL: Private Photo Dataset Protection by Data Poisoning for Vision-Language Models

Recent advances in visual-language alignment have endowed vision-language models (VLMs) with fine-grained image understanding capabilities. However, this progress also introduces new privacy risks. This paper first proposes a novel privacy threat model named identity-affiliation learning: an attacker fine-tunes a VLM using only a few private photos of a target individual, thereby embedding associations between the target facial identity and their private property and social relationships into the model's internal representations. Once deployed via public APIs, this model enables unauthorized exposure of the target user's private information upon input of their photos. To benchmark VLMs' susceptibility to such identity-affiliation leakage, we introduce the first identity-affiliation dataset comprising seven typical scenarios appearing in private photos. Each scenario is instantiated with multiple identity-centered photo-description pairs. Experimental results demonstrate that mainstream VLMs like LLaVA, Qwen-VL, and MiniGPT-v2, can recognize facial identities and infer identity-affiliation relationships by fine-tuning on small-scale private photographic dataset, and even on synthetically generated datasets. To mitigate this privacy risk, we propose DP2-VL, the first Dataset Protection framework for private photos that leverages Data Poisoning. Though optimizing imperceptible perturbations by pushing the original representations toward an antithetical region, DP2-VL induces a dataset-level shift in the embedding space of VLMs'encoders. This shift separates protected images from clean inference images, causing fine-tuning on the protected set to overfit. Extensive experiments demonstrate that DP2-VL achieves strong generalization across models, robustness to diverse post-processing operations, and consistent effectiveness across varying protection ratios.

cs.CV↗
arXiv2026

AMIF: Authorizable Medical Image Fusion Model with Built-in Authentication

Multimodal image fusion enables precise lesion localization and characterization for accurate diagnosis, thereby strengthening clinical decision-making and driving its growing prominence in medical imaging research. A powerful multimodal image fusion model relies on high-quality, clinically representative multimodal training data and a rigorously engineered model architecture. Therefore, the development of such professional radiomics models represents a collaborative achievement grounded in standardized acquisition, clinical-specific expertise, and algorithmic design proficiency, which necessitates protection of associated intellectual property rights. However, current multimodal image fusion models generate fused outputs without built-in mechanisms to safeguard intellectual property rights, inadvertently exposing proprietary model knowledge and sensitive training data through inference leakage. For example, malicious users can exploit fusion outputs and model distillation or other inference-based reverse engineering techniques to approximate the fusion performance of proprietary models. To address this issue, we propose AMIF, the first Authorizable Medical Image Fusion model with built-in authentication, which integrates authorization access control into the image fusion objective. For unauthorized usage, AMIF embeds explicit and visible copyright identifiers into fusion results. In contrast, high-quality fusion results are accessible upon successful key-based authentication.

cs.CV↗
arXiv2026

MetaKube: An Experience-Aware LLM Framework for Kubernetes Failure Diagnosis

Existing LLM-based Kubernetes diagnostic systems cannot learn from operational experience, operating on static knowledge bases without improving from past resolutions. We present MetaKube, an experience-aware LLM framework through three synergistic innovations: (1) an Episodic Pattern Memory Network (EPMN) that abstracts diagnostic patterns from historical resolutions and provides confidence-calibrated retrieval for both rapid pattern matching and guided causal exploration, (2) a meta-cognitive controller that dynamically routes between intuitive and analytical pathways based on problem familiarity, optimizing the trade-off between speed and depth, and (3) KubeLLM, a locally-deployable 8B model enhanced through domain-specific post-training on our 7,000-sample Kubernetes Fault Resolution Dataset. Evaluation on 1,873 real-world scenarios demonstrates MetaKube transforms Qwen3-8B from 50.9 to 90.5 points, approaching GPT-4.1 performance while ensuring complete data privacy. EPMN contributes 15.3% improvement through experiential learning, with continuous learning experiments showing progressive gains as the system accumulates operational knowledge. The source code and related resources are available at https://github.com/MetaKube-LLM-for-Kubernetes-Diagnosis/MetaKube.

cs.LG↗
arXiv2026

A Model Consistency-Based Countermeasure to GAN-Based Data Poisoning Attack in Federated Learning

In federated learning (FL), although the original intention of available but not visible data is to allay data privacy concerns, it potentially brings new security threats, particularly poisoning attacks that target such not visible local data. Intuitively, such data poisoning attacks have great potential in stealthily degrading global FL outcomes, and are expected to be even stealthier if being enhanced by generative models like generative adversarial networks (GANs). However, existing defense methods have not been thoroughly challenged in this regard and generally fail to be aware of a local generation of seemingly legitimate poisoned data. With a growing concern on potentially stealthier attacks, in this paper, a cost-effective defense mechanism named Model Consistency-Based Defense (MCD) is proposed, which offers a comprehensive examination of available local models across multiple feature dimensions, providing an indirect yet effective means of identifying hidden data poisoning attackers. To push the limit of MCD against stealthier attacks, we propose a new GAN-based data poisoning attack model named VagueGAN and an unsupervised variant of it, which can be flexibly deployed to generate seemingly legitimate but noisy poisoned data. The consistency of GAN outputs revealed by VagueGAN helps strengthen MCD to work against stealthier GAN-based attacks as well as other mainstream ones. Extensive experiments on multiple open datasets (MNIST, Fashion-MNIST, CIFAR-10, CIFAR-100, and Mini-Imagenet) indicate that our attack method better balances the trade-off between attack effectiveness and stealthiness with low complexity. More importantly, our defense mechanism is shown to be more competent in identifying a variety of poisoned data, particularly stealthier GAN-poisoned ones.

cs.CR↗
arXiv2026

Contrastive Retrieval Heads Improve Attention-Based Re-Ranking

The strong zero-shot and long-context capabilities of recent Large Language Models (LLMs) have paved the way for highly effective re-ranking systems. Attention-based re-rankers leverage attention weights from transformer heads to produce relevance scores, but not all heads are created equally: many contribute noise and redundancy, thus limiting performance. To address this, we introduce CoRe heads, a small set of retrieval heads identified via a contrastive scoring metric that explicitly rewards high attention heads that correlate with relevant documents, while downplaying nodes with higher attention that correlate with irrelevant documents. This relative ranking criterion isolates the most discriminative heads for re-ranking and yields a state-of-the-art list-wise re-ranker. Extensive experiments with three LLMs show that aggregated signals from CoRe heads, constituting less than 1% of all heads, substantially improve re-ranking accuracy over strong baselines. We further find that CoRe heads are concentrated in middle layers, and pruning the computation of final 50% of model layers preserves accuracy while significantly reducing inference time and memory usage.

cs.IR↗
arXiv2026

Accurate B meson and Bottomonium masses and decay constants from the tadpole improved clover ensembles

We present a determination of the bottom quark mass, the masses of S-wave bottom mesons, and their decay constants using an anisotropic clover fermion discretization for the heavy quark, on $2+1$ flavor isotropic QCD ensembles. Our analysis is based on 16 ensembles spanning 6 lattice spacings, with pion masses in the range of 135-350 MeV and several values of the strange quark mass. We demonstrate that the effective anisotropy parameter for the heavy quark approaches unity with controllable $\mathcal{O}(a^2)$ corrections. A non-perturbative renormalization procedure is developed and validated through predictions of the bottom quark mass and decay constants. This framework enables calculations at the physical $b$-quark mass even on lattices with spacing $a \sim 0.1$ fm, where $m_b a \sim 2.5$, while keeping discretization errors in hadronic matrix elements at the $\sim 10$% level which can be eliminated properly through the continuum extrapolation. Using the physical $Υ$ mass as input, we obtain $m_b^{\overline{\mathrm{MS}}}(m_b) = 4.185(37)$ GeV and the full spectrum of S-wave bottom mesons with 0.1% uncertainty or less. Pseudoscalar and vector decay constants and their ratios for all kinds of S-wave bottom mesons are also provided.

hep-lat↗
arXiv2026

LifeEval: A Multimodal Benchmark for Assistive AI in Egocentric Daily Life Tasks

The rapid progress of Multimodal Large Language Models (MLLMs) marks a significant step toward artificial general intelligence, offering great potential for augmenting human capabilities. However, their ability to provide effective assistance in dynamic, real-world environments remains largely underexplored. Existing video benchmarks predominantly assess passive understanding through retrospective analysis or isolated perception tasks, failing to capture the interactive and adaptive nature of real-time user assistance. To bridge this gap, we introduce LifeEval, a multimodal benchmark designed to evaluate real-time, task-oriented human-AI collaboration in daily life from an egocentric perspective. LifeEval emphasizes three key aspects: task-oriented holistic evaluation, egocentric real-time perception from continuous first-person streams, and human-assistant collaborative interaction through natural dialogues. Constructed via a rigorous annotation pipeline, the benchmark comprises 4,075 high-quality question-answer pairs across 6 core capability dimensions. Extensive evaluations of 26 state-of-the-art MLLMs on LifeEval reveal substantial challenges in achieving timely, effective and adaptive interaction, highlighting essential directions for advancing human-centered interactive intelligence.

cs.AI↗
arXiv2026

Don't let the information slip away

Real-time object detection has advanced rapidly in recent years. The YOLO series of detectors is among the most well-known CNN-based object detection models and cannot be overlooked. The latest version, YOLOv26, was recently released, while YOLOv12 achieved state-of-the-art (SOTA) performance with 55.2 mAP on the COCO val2017 dataset. Meanwhile, transformer-based object detection models, also known as DEtection TRansformer (DETR), have demonstrated impressive performance. RT-DETR is an outstanding model that outperformed the YOLO series in both speed and accuracy when it was released. Its successor, RT-DETRv2, achieved 53.4 mAP on the COCO val2017 dataset. However, despite their remarkable performance, all these models let information to slip away. They primarily focus on the features of foreground objects while neglecting the contextual information provided by the background. We believe that background information can significantly aid object detection tasks. For example, cars are more likely to appear on roads rather than in offices, while wild animals are more likely to be found in forests or remote areas rather than on busy streets. To address this gap, we propose an object detection model called Association DETR, which achieves state-of-the-art results compared to other object detection models on the COCO val2017 dataset.

cs.CV↗
arXiv2026

DM0: An Embodied-Native Vision-Language-Action Model towards Physical AI

Moving beyond the traditional paradigm of adapting internet-pretrained models to physical tasks, we present DM0, an Embodied-Native Vision-Language-Action (VLA) framework designed for Physical AI. Unlike approaches that treat physical grounding as a fine-tuning afterthought, DM0 unifies embodied manipulation and navigation by learning from heterogeneous data sources from the onset. Our methodology follows a comprehensive three-stage pipeline: Pretraining, Mid-Training, and Post-Training. First, we conduct large-scale unified pretraining on the Vision-Language Model (VLM) using diverse corpora--seamlessly integrating web text, autonomous driving scenarios, and embodied interaction logs-to jointly acquire semantic knowledge and physical priors. Subsequently, we build a flow-matching action expert atop the VLM. To reconcile high-level reasoning with low-level control, DM0 employs a hybrid training strategy: for embodied data, gradients from the action expert are not backpropagated to the VLM to preserve generalized representations, while the VLM remains trainable on non-embodied data. Furthermore, we introduce an Embodied Spatial Scaffolding strategy to construct spatial Chain-of-Thought (CoT) reasoning, effectively constraining the action solution space. Experiments on the RoboChallenge benchmark demonstrate that DM0 achieves state-of-the-art performance in both Specialist and Generalist settings on Table30.

cs.RO↗
arXiv2026

Deep Learning-Based Object Pose Estimation: A Comprehensive Survey

Object pose estimation is a fundamental computer vision problem with broad applications in augmented reality and robotics. Over the past decade, deep learning models, due to their superior accuracy and robustness, have increasingly supplanted conventional algorithms reliant on engineered point pair features. Nevertheless, several challenges persist in contemporary methods, including their dependency on labeled training data, model compactness, robustness under challenging conditions, and their ability to generalize to novel unseen objects. A recent survey discussing the progress made on different aspects of this area, outstanding challenges, and promising future directions, is missing. To fill this gap, we discuss the recent advances in deep learning-based object pose estimation, covering all three formulations of the problem, \emph{i.e.}, instance-level, category-level, and unseen object pose estimation. Our survey also covers multiple input data modalities, degrees-of-freedom of output poses, object properties, and downstream tasks, providing the readers with a holistic understanding of this field. Additionally, it discusses training paradigms of different domains, inference modes, application areas, evaluation metrics, and benchmark datasets, as well as reports the performance of current state-of-the-art methods on these benchmarks, thereby facilitating the readers in selecting the most suitable method for their application. Finally, the survey identifies key challenges, reviews the prevailing trends along with their pros and cons, and identifies promising directions for future research. We also keep tracing the latest works at https://github.com/CNJianLiu/Awesome-Object-Pose-Estimation.

cs.CV↗
arXiv2026

Surveillance Facial Image Quality Assessment: A Multi-dimensional Dataset and Lightweight Model

Surveillance facial images are often captured under unconstrained conditions, resulting in severe quality degradation due to factors such as low resolution, motion blur, occlusion, and poor lighting. Although recent face restoration techniques applied to surveillance cameras can significantly enhance visual quality, they often compromise fidelity (i.e., identity-preserving features), which directly conflicts with the primary objective of surveillance images -- reliable identity verification. Existing facial image quality assessment (FIQA) predominantly focus on either visual quality or recognition-oriented evaluation, thereby failing to jointly address visual quality and fidelity, which are critical for surveillance applications. To bridge this gap, we propose the first comprehensive study on surveillance facial image quality assessment (SFIQA), targeting the unique challenges inherent to surveillance scenarios. Specifically, we first construct SFIQA-Bench, a multi-dimensional quality assessment benchmark for surveillance facial images, which consists of 5,004 surveillance facial images captured by three widely deployed surveillance cameras in real-world scenarios. A subjective experiment is conducted to collect six dimensional quality ratings, including noise, sharpness, colorfulness, contrast, fidelity and overall quality, covering the key aspects of SFIQA. Furthermore, we propose SFIQA-Assessor, a lightweight multi-task FIQA model that jointly exploits complementary facial views through cross-view feature interaction, and employs learnable task tokens to guide the unified regression of multiple quality dimensions. The experiment results on the proposed dataset show that our method achieves the best performance compared with the state-of-the-art general image quality assessment (IQA) and FIQA methods, validating its effectiveness for real-world surveillance applications.

eess.IV↗
arXiv2026

VQAThinker: Exploring Generalizable and Explainable Video Quality Assessment via Reinforcement Learning

Video quality assessment (VQA) aims to objectively quantify perceptual quality degradation in alignment with human visual perception. Despite recent advances, existing VQA models still suffer from two critical limitations: \textit{poor generalization to out-of-distribution (OOD) videos} and \textit{limited explainability}, which restrict their applicability in real-world scenarios. To address these challenges, we propose \textbf{VQAThinker}, a reasoning-based VQA framework that leverages large multimodal models (LMMs) with reinforcement learning to jointly model video quality understanding and scoring, emulating human perceptual decision-making. Specifically, we adopt group relative policy optimization (GRPO), a rule-guided reinforcement learning algorithm that enables reasoning over video quality under score-level supervision, and introduce three VQA-specific rewards: (1) a \textbf{bell-shaped regression reward} that increases rapidly as the prediction error decreases and becomes progressively less sensitive near the ground truth; (2) a \textbf{pairwise ranking reward} that guides the model to correctly determine the relative quality between video pairs; and (3) a \textbf{temporal consistency reward} that encourages the model to prefer temporally coherent videos over their perturbed counterparts. Extensive experiments demonstrate that VQAThinker achieves state-of-the-art performance on both in-domain and OOD VQA benchmarks, showing strong generalization for video quality scoring. Furthermore, evaluations on video quality understanding tasks validate its superiority in distortion attribution and quality description compared to existing explainable VQA models and LMMs. These findings demonstrate that reinforcement learning offers an effective pathway toward building generalizable and explainable VQA models solely with score-level supervision.

cs.CV↗
arXiv2026

Mitigating Data Centers Load Risks and Enabling Grid Support Functions through Grid-Forming Control

The rapid growth of hyperscale data centers driven by Large Language Models and Artificial Intelligence workloads has introduced new challenges for power systems. These facilities experience abrupt power variations during model training and check-point-saving events, causing voltage deviations and frequency disturbances. Moreover, they operate as passive loads that draw power without offering any grid support. This paper presents an integrated architecture that combines Battery Energy Storage Systems (BESSs) within data centers using Grid-Forming inverters to provide active grid-support functions. Simulation results through MATLAB/Simulink demonstrate accurate power reference tracking under dynamic loading, with eight coordinated BESS units supplying instantaneous power during training and saving conditions. Under single-phase voltage depression near the data center bus, the BESS delivered reactive power support similar to a Static Synchronous Compensator. During grid disconnection, seamless islanded operation was achieved with stable voltage, frequency, and continuous power delivery at the data center bus.

eess.SY↗
arXiv2026

SpyDir: Spy Device Localization Through Accurate Direction Finding

Hidden spy cameras have become a great privacy threat recently, as these low-cost, low-power, and small form-factor IoT devices can quietly monitor human activities in the indoor environment without generating any side-channel information. As such, it is difficult to detect and even more challenging to localize them in the rich-scattering indoor environment. To this end, this paper presents the design, implementation, and evaluation of SpyDir, a system that can accurately localize the hidden spy IoT devices by harnessing the electromagnetic emanations automatically and unintentionally emitted from them. Our system design mainly consists of a portable switching antenna array to sniff the spectrum-spread emanations, an emanation enhancement algorithm through non-coherent averaging that can de-correlate the correlated noise effect due to the square-wave emanation structure, and a multipath-resolving algorithm that can exploit the relative channels using a novel optimization-based sparse AoA derivation. Our real-world experimental evaluation across different indoor environments demonstrates an average AoA error of 6.30 deg, whereas the baseline algorithm yields 21.06 deg, achieving over a 3.3 times improvement in accuracy, and a mean localization error of 19.86cm over baseline algorithms of 206.79cm (MUSIC) and 294.75cm (SpotFi), achieving over a 10.41 times and 14.8 times improvement in accuracy.

cs.CR↗
arXiv2026

Physics Informed Reconstruction of Four-Dimensional Atmospheric Wind Fields Using Multi-UAS Swarm Observations in a Synthetic Turbulent Environment

Accurate reconstruction of atmospheric wind fields is essential for applications such as weather forecasting, hazard prediction, and wind energy assessment, yet conventional instruments leave spatio-temporal gaps within the lower atmospheric boundary layer. Unmanned aircraft systems (UAS) provide flexible in situ measurements, but individual platforms sample wind only along their flight trajectories, limiting full wind-field recovery. This study presents a framework for reconstructing four-dimensional atmospheric wind fields using measurements obtained from a coordinated UAS swarm. A synthetic turbulence environment and high-fidelity multirotor simulation are used to generate training and evaluation data. Local wind components are estimated from UAS dynamics using a bidirectional long short-term memory network (Bi-LSTM) and assimilated into a physics-informed neural network (PINN) to reconstruct a continuous wind field in space and time. For local wind estimation, the bidirectional LSTM achieves root-mean-square errors (RMSE) of 0.064 and 0.062 m/s for the north and east components in low-wind conditions, increasing to 0.122 to 0.129 m/s under moderate winds and 0.271 to 0.273 m/s in high-wind conditions, while the vertical component exhibits higher error, with RMSE values of 0.029 to 0.091 m/s. The physics-informed reconstruction recovers the dominant spatial and temporal structure of the wind field up to 1000 m altitude while preserving mean flow direction and vertical shear. Under moderate wind conditions, the reconstructed mean wind field achieves an overall RMSE between 0.118 and 0.154 m/s across evaluated UAS configurations, with the lowest error obtained using a five-UAS swarm. These results demonstrate that coordinated UAS measurements enable accurate and scalable four-dimensional wind-field reconstruction without dedicated wind sensors or fixed infrastructure.

cs.LG↗