Searcharxiv⌕ Search

arXiv subjects

Kai Wang

Publications and source records attributed to Kai Wang.

At least 19 recordsLinked to original sources
arXiv2026

Mixed-Type Relational Database Synthesis with Graph-Prompted Language Model

Data access is often limited by privacy or proprietary restrictions, motivating relational database synthesis, which synthesizes realistic contents across related tables. Relational databases couple multi-table structure with mixed-type attributes and intricate semantic dependencies among values, making realistic synthesis challenging. In this paper, we propose REMIX, a unified framework for realistic relational database synthesis with mixed-type attributes. REMIX integrates a small language model (SLM) and a large language model (LLM) through two key phases: Graph-to-Prompt Tuning and Collaborative SLM-LLM Decoding. To capture attribute dependencies within and across tables, Graph-to-Prompt Tuning represents the database as an attributed heterogeneous graph. We project mixed-type record embeddings and aggregated relational context into soft prompts that serve as continuous vectors for an SLM. We jointly optimize the graph-to-prompt mapping and SLM through autoregressive attribute reconstruction. Furthermore, to reduce semantic inconsistencies, Collaborative SLM-LLM Decoding combines SLM-based record generation with repair programs induced by an LLM. The LLM discovers semantic dependencies, iteratively refines them through independent validation with a statistical bound on the overall probability of false acceptance, and translates validated dependencies into tested programs that repair generated values. Extensive experiments on seven real-world relational databases demonstrate the effectiveness of REMIX in generating high-fidelity data with improved semantic consistency and downstream utility.

cs.DB↗
arXiv2026

Small Hankel operators on weighted Besov spaces

We characterize bounded small Hankel operators from $B_p(\om)$ to $\overline{B_q(\om)}$ on the polydisc for $0<p,q<\infty$, under explicit conditions relating the kernel parameter to the weight. When $q<p$, boundedness is equivalent to a weighted Besov condition on the symbol; when $p\le q$, it is equivalent to a weighted supremum condition on a shifted radial derivative. In particular, the diagonal case is characterized by the product Bloch space. The proof combines weighted Bergman duality and atomic factorization with local kernel estimates and randomized atomic tests.

math.FA↗
arXiv2026

Encode Once, Answer Many: Internalizing Context via Parametric Flow

Large language models are typically adapted to a task either by placing knowledge in the input context, which keeps it external and re-encodes it on every call, or by fine-tuning, which internalizes it but demands a costly round of optimization per task. We aim to internalize the knowledge carried by a context directly into model parameters, without per-task optimization and without keeping the context in the input at inference time. To this end, we propose CLIK, a context-conditioned parameter generator that maps a given context to LoRA parameters for internalizing knowledge in a single forward pass through a Transformer encoder with cross-attention, requiring no test-time gradient updates. Once the generated LoRA is inserted into a frozen base model, the model answers downstream queries from the query alone, having internalized the context into its weights rather than re-reading it. Because the adapter depends only on the context, it is generated once and reused for every query on the same document, so the cost of reading that document is amortized across queries. On document question answering, CLIK substantially improves over the no-context base model and outperforms existing hypernetwork-based generators, while remaining behind a base model that reads the context directly, and it carries over to multi-turn dialogue without any multi-turn-specific training. Analysis further shows that the context-dependent part of the generated adapter concentrates in a compact and specific substructure.

cs.AI↗
arXiv2026

Trustworthy Execution of Industrial Agents

Industrial agents increasingly modify persistent engineering objects rather than merely generate answers. Authentication and successful calls cannot establish whether intended changes occurred, evidence supports a conclusion, or historical results remain applicable. We present a framework for trustworthy industrial-agent execution within a single enterprise governance domain. Eight control domains assign responsibilities across authority, planning, composition, tool execution, actual effects, evidence, baseline and use governance, and recovery and evolution. A four-plane architecture and three control chains connect these responsibilities through typed objects and version-bound interfaces. The specification separates tool returns from actual engineering effects, evidence coverage from engineering judgment, and historical lifecycle state from current applicability. Multi-agent handoffs are an architectural extension examined through deterministic interface tests. Distinct prototypes examine selected mechanisms using conformance tests, ablations, uncertainty stress, CAD/CAE execution, and local persistence. An OpenFOAM fixture produces COMPLETE coverage with FAIL judgment; an expired-job case requires readmission despite completed computation. CAD and recovery counterexamples expose omitted mandatory checks, discarded negative facts, and inconsistent producer-consumer mappings. Targeted repairs support local regression conclusions. The reported 0/29 hazardous commits measure closed-set specification conformance, not a safety probability. We distinguish the complete framework, implemented paths, and empirical evidence, identifying check-inventory completeness and cross-module fact preservation as prerequisites for extending these results.

cs.SE↗
arXiv2026

ParaBlock++: Communication-Computation Parallel Blockwise Hierarchical Federated Learning

In the emerging 6G paradigm, where AI-driven services operate over decentralized and resource-constrained infrastructures, the convergence of distributed learning algorithms remains a significant challenge. This paper introduces $\mathrm{ParaBlock}$++, a low Earth orbit (LEO) satellite-assisted hierarchical federated learning (HFL) framework, enabling resource-limited terrestrial clients to collaboratively engage in federated learning (FL) via ground-satellite links (GSLs) supported by LEO satellite megaconstellations. Within this architecture, LEO satellites serve multiple roles: as edge servers aggregating and broadcasting models, as clients accelerating training, and as high-throughput relays managing inter-satellite routing through inter-satellite links (ISLs). Departing from conventional synchronous Jacobian-type updates, $\mathrm{ParaBlock}$++ employs asynchronous BCD-type updates to enhance convergence speed. We provide theoretical convergence guarantees for convex, strongly convex, and non-convex objectives, and formulate optimization problems to determine optimal cluster resource allocation and model partitioning for minimizing global training latency. Experimental evaluations demonstrate that $\mathrm{ParaBlock}$++ achieves faster initial convergence and comparable final performance for deep neural networks (DNNs) such as ResNet-18, offering promising directions for federated learning across geographically remote areas enabled by 6G technologies, but much less global rounds and communication cost.

cs.LG↗
arXiv2026

Hierarchical Asynchronous Federated Learning over Space-Air-Ground Integrated Networks

With the emerging 6G paradigm where AI services operate over decentralized and resource-constrained edge infrastructure, performance of distributed learning algorithms remains a significant challenge. This paper proposes HAFL, a Low Earth Orbit (LEO) satellite-assisted hierarchical asynchronous Federated Learning (FL) framework, enabling resource-limited terrestrial clients to collaborate via Ground-Satellite Links (GSLs). Within this framework, LEO satellites serve multiple roles: as edge servers aggregating and broadcasting models, as clients accelerating training, and as high-throughput relays managing inter-satellite routing through Inter-Satellite Links (ISLs). Unlike conventional synchronous Jacobian-type updates, HAFL employs asynchronous Gauss-Seidel-type updates to enhance convergence speed. We provide theoretical convergences for convex, strongly convex, and nonconvex objectives, respectively, and formulate optimization problems to jointly determine optimal hyperparameters to minimize global training latency. Evaluations demonstrate that HAFL achieves faster initial convergence and comparable final performance for benchmark DNNs and datasets, offering promising FL systems across remote areas with 6G technologies, but much less global rounds and communication cost.

cs.LG↗
arXiv2026

Who Gets Watched? Allocating Scarce Escalation Resources on Pediatric Wards with Neural Index Policies

Problem definition: Hospitals use separate pathways to identify and respond to deteriorating ward patients. At Children's Healthcare of Atlanta (CHOA), these include Watcher surveillance, Critical Care Outreach Team assessment, and transfer to the intensive care unit. When several patients need attention simultaneously, however, no common process determines who should receive which response under limited staffing and capacity. We study how to coordinate these actions to support earlier escalation without unnecessary monitoring. Methodology/results: We model ward escalation decisions using a multi-action restless multi-armed bandit model and develop Multi-action Index Learning and Optimization Policy (MILOP) as a computational solution approach. MILOP trains a neural network to generate a high-quality index policy using a decision-focused learning approach. We train and evaluate MILOP using historical CHOA data and compare it with two practical baselines. MILOP recommends an escalation before 87.4% of critical or emergent intensive-care transfers, compared with 79.5% for the stronger baseline, while using 13.5% less monitoring resources. MILOP provides a median of 19.2 hours lead time before transfer, earlier than either baseline. Moreover, 99.3% of MILOP's proactive transfer recommendations are preceded by Watcher surveillance or outreach-team assessment, demonstrating a desirable escalation of care. Managerial implications: MILOP can help leaders focus scarce clinical attention, gain more time to respond before severe deterioration, and reduce monitoring burden compared to existing practices. Broadly, hospitals may improve escalation decisions by coordinating existing pathways rather than managing each separately or using risk scores as stand-alone decision rules.

cs.LG↗
arXiv2026

First-Order Bilevel Optimization for Non-truthful Auction Design

Automated mechanism design has enabled the learning of complex auction mechanisms using neural networks. However, designing non-truthful auctions remains challenging because bidders' equilibrium strategies depend on the auction mechanism and thus evolve as the mechanism is optimized. In this work, we formulate auction design as a Stackelberg game with multiple strategic followers and develop a fully first-order algorithm for revenue optimization in non-truthful auctions. We establish non-asymptotic guarantees for hypergradient approximation and show that, under standard smoothness and strong-monotonicity assumptions, our method reaches an $O(ε)$-stationary point using $\tilde{\mathcal{O}}(ε^{-2})$ first-order oracle calls. Experiments on symmetric and asymmetric single-item auctions, as well as multi-item auctions, show that our method converges faster than second- and zeroth-order baselines and scales better as the number of bidders increases. Our method also discovers auctions that achieve revenue close to truthful baselines. These results provide a scalable framework for automated mechanism design beyond truthful auctions.

cs.GT↗
arXiv2026

Deformation of nef adjoint canonical line bundles

Much inspired by J. A. Wiśniewski's nef-value function method, we prove that in a smooth projective family over the unit disk, if the adjoint bundle of the canonical line bundle with a relatively semiample line bundle is nef on one fiber, then it remains nef on all fibers. We further extend this result to the semiampleness of the adjoint canonical line bundles. Using these, we prove the deformation invariance of any generalized plurigenera by assuming that only one fiber admits the semiample canonical line bundle and improve the first author--Xiao-Lei Liu's recent deformation rigidity of projective manifolds with semiample canonical line bundles. In particular, also by E. Viehweg--K. Zuo's result on the minimal number of singular fibers in a family and the first author--X. Liu's isotriviality result, if a projective family over $\mathbb{P}^1$ or an elliptic curve has one fiber with the big and nef (or more generally semiample) canonical line bundle, then all fibers are isomorphic to this fiber. Next, much inspired by M. Andreatta--T. Peternell's deformation theoretical approach, we prove that, in a smooth Kähler family of threefolds, if the canonical line bundle of one fiber is not nef, then none of its small deformations admits a nef canonical line bundle either. This partially confirms a problem posed by F. Campana--T. Peternell and the global stability of semiampleness of canonical line bundles of threefolds under a Kähler smooth deformation.

math.AG↗
arXiv2026

AWARE: Action-Entropy-Weighted and Attention-Guided Retrieval of Observation Evidence for Vision-Language-Action Models

Vision-Language-Action (VLA) models combine multimodal perception with robotic control, demonstrating remarkable potential for generalizable manipulation. Yet, enhancing their execution typically relies on auxiliary sensory inputs (e.g., depth, 3D point clouds) or costly fine-tuning. Inspired by the insight that Feed-Forward Networks (FFNs) act as key-value memory, we propose Action-Entropy-Weighted and Attention-Guided Retrieval of Observation Evidence (AWARE), an effective, training-free, and plug-and-play module for VLAs. Specifically, AWARE tracks layer-wise uncertainty via an element-wise Action Entropy metric, and retrieves observation evidence into the element's FFN through associative memory lookup. Crucially, the retrieved features are weighted by instruction-guided cross-modal attention to isolate task-relevant observation cues, while injection strength is adaptively scaled by the element's entropy and relative network depth. This selectively channels observation evidence to high-entropy action dimensions and deep layers where grounding decays, while leaving converged representations intact. Across LIBERO, SIMPLER, CALVIN, and real-world robot manipulation, AWARE consistently improves diverse VLA architectures with minimal computational overhead, achieving the largest gains on long-horizon tasks without architectural changes or additional training. Project page: https://jiabingyang01.github.io/AWARE/.

cs.CV↗
arXiv2026

Spectral fingerprint recognition of mobile signals via Rydberg atomic sensors

Rydberg-atom sensors have emerged as a promising quantum technology for broadband radio-frequency detection owing to their intrinsic calibration, wide frequency coverage, and high sensitivity. However, their potential for intelligent identification of real-world wireless signals remains largely unexplored. Here, we demonstrate a quantum-enabled framework for mobile signal classification based on Rydberg electromagnetically induced transparency spectroscopy and deep learning. Three types of wireless emissions, including cellular voice call, Wi-Fi, and 5G signals from three commercial smartphone brands, are measured to construct a spectral-fingerprint dataset. A convolutional neural network trained on this dataset achieves an initial baseline classification accuracy of 93.44\%. To compensate for spectral drifts induced by temporal and environmental variations, a progressive fine-tuning strategy is employed to continuously adapt the pretrained model, enabling robust classification across diverse deployment scenarios. Furthermore, continuous real-time monitoring is demonstrated under dynamically varying wireless conditions, including cellular voice call, Wi-Fi, and their coexistence. The proposed system achieves an overall frame-level recognition accuracy of 87.67\% over a 60-s measurement. These results establish a pathway toward intelligent quantum radio-frequency sensing and highlight the potential of Rydberg-atom platforms for real-time wireless signal awareness and spectrum monitoring.

quant-ph↗
arXiv2026

Rydberg atomic field-enhanced feed beyond electrically small antennas

The reception of electromagnetic signals becomes increasingly challenging at longer wavelengths, as conventional antenna feeds must scale with the operating wavelength, making it difficult to reconcile compactness with high sensitivity. Rydberg atoms enable direct field-to-optical conversion within deeply subwavelength sensing volumes, yet an effective receiving feed that couples incident fields into the atomic interaction volume remains underdeveloped. Here, we introduce a Rydberg atomic field-enhanced feed (RAFEF) that integrates electromagnetic field capture and enhancement with Rydberg-atomic transduction. We establish a physical model to quantify its receiving capability, Phi, that jointly accounts for sensitivity enhancement, receiving bandwidth and normalized three-dimensional volume, and experimentally obtain Phi_dB = 70.81 dB. Compared with a conventional loop-antenna feed, the RAFEF provides a higher received SNR and overall receiving capability, while its tunable narrowband response enables frequency selectivity directly at the receiving-feed stage. We further demonstrate selective recovery of a target channel from simultaneously incident signals and reception of a 93.5 MHz FM broadcast after approximately 10 km of over-the-air propagation. These results establish the RAFEF as an atomic receiving-feed architecture that exploits localized electric-field transduction to extend Rydberg-atomic reception beyond conventional electrically small antenna feeds.

physics.atom-ph↗
arXiv2026

Grothendieck rings of near-Tambara--Yamagami fusion categories

We determine the Grothendieck rings of fusion categories over $\mathcal{C}$ containing a nonpointed Tambara--Yamagami fusion subcategory and exactly one further simple object. The subcategory necessarily has rank five, its invertible objects form the Klein four-group, and the coefficient $κ$ of the additional simple object in its tensor square belongs to $\{0,2,7\}$. Each of these three based rings is categorifiable. For $κ=0$, every category belongs, up to tensor equivalence, to the rank-one odd-component extension list of Choi--Lu--Sun. For $κ=2$ we construct an explicit unitary group-theoretical realization, and for $κ=7$ a realization is $\rm{Rep}(\rm{PSU}(3,2))$.

math.QA↗
arXiv2026

LaWAM: Latent World Action Models for Efficient Dynamics-Aware Robot Policies

Vision-Language-Action models (VLAs) leverage large-scale vision-language pretraining for semantic robot control, but often lack explicit foresight into how robot actions change the scene. World-Action Models (WAMs) address this limitation by conditioning policies on predicted futures, yet existing approaches typically rely on computationally expensive video generation with substantial pixel-level redundancy. We present LaWAM, a Latent World Action Model that exposes predictive dynamics to robot policies through compact latent visual subgoals instead of reconstructed future video. At the core of LaWAM is a latent-action-conditioned Latent World Model (LaWM). We obtain LaWM by training a latent action model in the latent space of a pretrained vision foundation model and repurposing its forward decoder to predict future observation features for scene evolution. LaWAM then conditions action generation on these predicted latent visual subgoals to enable dynamics-aware robot control. LaWAM achieves state-of-the-art or competitive success rates (SRs) across LIBERO (98.6% SR), RoboTwin (91.22% SR), and real-world manipulation tasks while retaining low-latency inference. LaWAM runs in 187 ms per action-chunk prediction and achieves up to 24x lower wall-clock latency than pixel-space WAMs.

cs.RO↗
arXiv2026

Coupling Execution Horizon and Sampling Depth in Flow-Matching VLA Policies

Flow-matching Vision-Language-Action (VLA) models generate expressive continuous trajectories for robotic manipulation, yet balancing real-time responsiveness with execution stability remains a persistent challenge. Standard deployment workflows typically treat numerical solver depth and execution horizon as decoupled choices. The horizon amortizes inference latency, while solver depth is kept high to preserve integration fidelity. In this paper, we reveal a fundamental compute-execution coupling in flow-matching VLAs governed by a systematic sign reversal. While coarse numerical integration degrades precision under rapid replanning, it markedly improves task performance under extended open-loop commitment. Geometrically, numerical truncation contracts lateral action dispersion around an invariant trajectory centroid, serving as an implicit regularizer that suppresses compounding open-loop drift. Across LIBERO, RoboTwin 2.0, and real-world Franka manipulation on representative VLA backbones including $π_0$, $π_{0.5}$, and GR00T, exploiting this joint coupling consistently achieves superior task success while drastically lowering inference latency, entirely training-free and without architectural modifications. These findings establish numerical sampling depth as an operational variance regulator, shifting its role from a static precision budget to an active deployment knob in robot learning. Project page: https://jiabingyang01.github.io/Horizon-Depth-VLA/

cs.RO↗
arXiv2026

The Maximal Robust Positively Invariant Set for Linear Difference Inclusions

This article develops a systematic characterization and exact computational framework for the maximal robust positively invariant set of linear difference inclusions. The maximal set is characterized as the greatest fixed point of the robust predecessor operator, leading to three equivalent set iterations for its computation. Under uniform exponential stability and bounded disturbances, the Kuratowski upper limit of the disturbance-reachable sets yields a necessary and sufficient condition for the maximal set to be nonempty under closed state constraints. Sufficient conditions for its finite determination are also established, together with an explicit bound on the determination index. For polyhedral state constraints, bounded disturbances, and finite or polytopic matrix families, exact half-space implementations of the three iterations are derived. Although these algorithms generate the same set sequence, complexity analysis and numerical experiments reveal substantial differences in their computational performance.

math.OC↗
arXiv2026

DiDE:Direct Injection with Color-Texture DEcoupling for 3D Stylization

Recent advances in rectified flow-based image-to-3D generative models have enabled high-fidelity 3D asset generation. Building on this, a growing line of work has exploited these strong 3D priors for training-free stylization, transferring visual attributes from a reference image onto a generated 3D asset. However, existing methods enforce an all-or-nothing paradigm: color and texture are transferred jointly, with no mechanism to control them independently -- a limitation we formalize as Disentangled 3D Stylization(Disen3D). To address this, we propose DiDE, the first training-free framework for Disen3D. Key to our approach is the observation that the structured latent space of image-to-3D models is overcomplete with respect to texture: texture information occupies only a small subset of the style-significant channels, leaving a free subspace available for independent color encoding. DiDE exploits this via a channel partition mechanism that processes a content image, a texture reference, and a color reference through dedicated branches and composes both style signals interference-free at every self-attention layer, preserving content geometry throughout. Experiments on Disen3D-Bench, our newly collected multi-reference benchmark, show that DiDE consistently outperforms 2D and 3D stylization baselines in color fidelity, texture transfer, and content preservation.

cs.CV↗
arXiv2026

Relational Geometry Forcing for World Action Models

World Action Models (WAMs) have shown strong potential for robot manipulation by jointly learning visual dynamics and executable actions. However, their video and action objectives do not explicitly constrain the 3D geometry that is crucial for precise manipulation. Recently, geometry-aware WAMs introduce depth or 4D prediction branches, whereas representation alignment methods supervise intermediate WAM features with pretrained 3D foundation models. Yet we find that such pointwise alignment benefits the video features but degrades action learning, since dense geometric feature targets overly restrict compact action features. To address this limitation, we propose Relational Geometry Forcing (RGF), which transfers geometric priors from a frozen 3D foundation model by aligning pairwise feature relations instead of pointwise features. Specifically, RGF applies three complementary relational objectives, where intra-frame and cross-view relations transfer scene geometry and multi-view correspondence to the video features, while action-temporal relations couple action dynamics to geometric scene evolution. Applied only during training, RGF requires no explicit geometry branch or extra inference module. Extensive experiments in simulation and the real world demonstrate the effectiveness of Relational Geometry Forcing, showing consistent gains across diverse manipulation scenarios and WAM paradigms.

cs.RO↗