arXiv subjects

Jing Huang

Publications and source records attributed to Jing Huang.

At least 19 recordsLinked to original sources

arXiv2026

DocSkill: Skill-Integrated Agentic Reasoning for Multilingual Multi-Hop Document Question Answering

Multilingual multi-hop document question answering (MMHop-DocQA) requires locating and integrating evidence across complex layouts, document pages, and diverse languages. Existing approaches include single-pass processing and adaptive tool use, but learning to coordinate evidence acquisition, cross-page integration, and answer verification remains challenging. To address this challenge, we present DocSkill, an agentic framework with a learned skill-orchestration policy for adaptive evidence acquisition and reasoning. The policy selects, revisits, or skips skills according to the evolving evidence state. To construct supervision for skill orchestration, bounded Monte Carlo Tree Search (MCTS) explores alternative skill trajectories offline. These trajectories provide demonstrations for supervised fine-tuning (SFT) and preference pairs for Direct Preference Optimization (DPO). We further introduce MMHop-DocBench, a multilingual multi-hop document QA dataset with training and test splits, containing 60K QA pairs across 10 languages and eight document domains. Its HopGen pipeline builds hierarchical page-element graphs to derive traceable cross-page evidence paths and generate grounded QA pairs with page- and element-level evidence annotations. Experiments show that DocSkill with Qwen3-VL-8B achieves 41.0 accuracy and 49.2 F1 on MMHop-DocBench, exceeding the strongest evaluated baseline by 5.5 and 6.3 points, respectively, and achieving competitive performance on MMLongBench-Doc and DocBench.

cs.CL↗

arXiv2026

Firefly: Illuminating Large-Scale Verified Tool-Call Data Generation from Real APIs

Training tool-calling agents requires large-scale trajectory data with verifiable labels, yet existing approaches either synthesize environments that diverge from real API behavior or generate tasks without ground-truth outcomes for verification. We present FireFly, a pipeline for generating verified tool-call data from real-world MCP servers. Our key insight is to invert the standard synthesis pipeline: rather than generating tasks and hoping they are solvable, we first let a strong LLM explore real APIs along graph-guided DAG structures, then synthesize tasks backward from observed outcomes, guaranteeing label correctness by construction. To handle the scale of real-world tool spaces (${\sim}$1,000 tools), we build a pairwise tool graph and sample sub-DAGs to focus exploration on semantically coherent workflows. To address environment drift in live APIs, we construct a retrieval-augmented simulator that caches all exploration results and replays them during training and evaluation, enabling fully offline and reproducible RL. Applying this pipeline yields 5,144 verified tasks spanning 240 servers and 993 tools. A 4B-parameter model trained with GRPO on FireFly matches Claude Sonnet 4.6 on our held-out test set and shows improvements on multiple tool-calling benchmarks including Tau2-Bench, MCPMark, and MCP-Atlas.

cs.SE↗

arXiv2026

Agents' Last Exam

Recent AI systems have achieved strong results on a wide range of benchmarks, yet these gains have not translated into economically meaningful deployment across many professional domains. We argue that this gap is largely an evaluation problem: widely used benchmarks lack sustained performance measurement on real and economically valuable workflows. This paper introduces Agents' Last Exam (ALE), a benchmark designed to evaluate AI agents on long horizon, economically valuable, real world tasks with verifiable outcomes. Developed in collaboration with 250+ industry experts, ALE covers non-physical industries defined with reference to O*NET / SOC 2018 (the U.S. federal occupational taxonomy). It is organized around a task taxonomy with 55 sub fields grouped into 13 industry clusters covering 1K+ tasks. Current results show that the hardest tier remains far from saturated: across mainstream harness and backbone configurations, the average full pass rate is below 1%. ALE is designed as a living benchmark: its task pool grows continuously as new workflows and industries are onboarded. More broadly, ALE is intended not merely as another leaderboard, but as an instrument for closing the gap between benchmark success and GDP relevant impact.

cs.AI↗

arXiv2026

The Golomb--Welch conjecture for lattice tilings

We prove the Golomb--Welch conjecture for lattice tilings: for every $n\ge3$ and $r\ge2$, the integer lattice $\Z^n$ admits no lattice tiling by Lee balls $B_n(r)$ of radius $r$. Together with the classical constructions for $n\le2$ and for $r=1$, this determines all parameters for which a lattice Lee tiling exists. Equivalently, we show that every abelian Cayley graph of even degree $2n\ge6$ and diameter $r\ge2$ has strictly fewer than $|B_n(r)|$ vertices.

cs.IT↗

arXiv2026

The Golomb--Welch conjecture for arbitrary centre sets

We prove that for all $n\ge3$ and $r\ge2$, the Lee ball $B_n(r)$ does not tile $\Z^n$ by translations, resolving the Golomb--Welch conjecture for arbitrary centre sets and completing the existence classification of perfect Lee codes in $\Z^n$. The proof combines translation-invariant averaging, boundary contact identities, and integer moment inequalities with coefficient comparisons in stable symmetric functions. Consequently, we obtain finite local obstructions, strict Banach density gaps for packings and coverings, and the nonexistence of exact measurable Lee-ball towers. In the lattice specialization, our result classifies the attainability of the Lee bound for even-degree abelian Cayley graphs, excludes normal topological coverings and Galois graph coverings for Lee-ball completions, and gives a strict node bound for lattice cubature as well as the nonexistence of subgroup-indexed orthogonal exponential bases on $B_n(r)$.

math.CO↗

arXiv2026

Deep holes of full-length Reed--Solomon codes

For every prime power $q>19683$, we settle the Cheng--Murray deep-hole problem for full-length Reed--Solomon codes in every dimension $2\le k\le q-2$, with the characteristic-two cases stated explicitly. For odd $q$, the deep holes are precisely the words of degree $k$. For even $q$ and $k\ge3$, the same conclusion holds except for an additional family when $k=q-3$. In even characteristic and dimension two, every deep hole has a unique expression $a h(X)+bX+c$, where $a\ne0$ and $h$ is a normalized $o$-polynomial. The proof combines quadratic quotient algebras, graph-theoretic weighted independent-set inequalities, and planar arc completion. We also obtain normal rational curve rigidity in the corresponding nonexceptional dimensions, the projective Reed--Solomon covering radii, and distinct-coordinate solvability and remainder criteria.

cs.IT↗

arXiv2026

On the MDS conjecture in even characteristic

We prove that every nontrivial linear $[n,k]$ maximum distance separable (MDS) code over a finite field $\F_q$ of characteristic two satisfies $n\le q+1$, except in dimensions $k=3$ and $k=q-1$, where $n\le q+2$. This resolves the linear MDS conjecture in even characteristic.

math.CO↗

arXiv2026

The Maximum Commute Time of Regular Graphs

For a connected simple regular graph $G$, let $τ(G)$ be the maximum commute time for simple random walk, and let $t_n$ be the maximum of $τ(G)$ over all such graphs of order $n$. Aldous and Fill conjectured that $t_n\sim3n^2/2$. We determine $t_n$ exactly and classify all maximizers for every even $n\geq10$ and every odd $n\geq15$; both thresholds are best possible. The maximizers are cubic block chains joined by bridges at even orders and quartic block chains joined at cut vertices at odd orders. The exact formulas specify the linear corrections according to the order modulo four and five, respectively, and imply $ t_n\sim \begin{cases} \dfrac32 n^2, & n\text{ even},\\[1mm] n^2, & n\text{ odd}, \end{cases} \qquad n\to\infty. $ Thus the conjectured asymptotic formula holds along even orders but fails along odd orders. The proof combines unit-current potential orderings, cut-current inequalities, and sharp block resistance bounds, supplemented by finite computations in exact rational arithmetic.

math.CO↗

arXiv2026

Design Anything, Design at Scale

As multimodal resources are transformed into media outputs, agentic AI is expected to stay aligned with design intent over long horizons. Designers follow briefs specifying goals, constraints, and deliverables, with later work building on earlier decisions. Recent work reuses design experience across tasks, yet alignment among artifacts within one project remains challenging. Agents therefore need to preserve and apply this intent across domains and as projects grow. We introduce Design Anything, a graph-based harness that preserves design intent outside the model and binds deliverables to the workflow steps that produce them through an intent-execution graph. A task contract retains constraints and decisions, selected designs become explicit inputs to later generations, and a deterministic controller uses bindings and execution records to check output types and counts, identify outstanding deliverables, and guide subsequent turns. We also introduce DesignAnything-Bench, with 228 tasks across 19 design domains, including 38 project families at scales of 6, 14, and 24 artifacts. With the LLM held fixed, Design Anything satisfies 85.0% and 92.5% of the brief's factual constraints with DeepSeek V4.1 Flash High and GPT-5.6 Sol High, respectively, exceeding the strongest same-model baselines by 15.4% and 14.4%. These gains reach 25.1% and 23.0% on 24-artifact tasks. Under each model, Design Anything leads or ties in constraint satisfaction in 18 of 19 domains. Professional designers also prefer its complete deliverable sets over those of same-model baselines in a blind evaluation.

cs.CV↗

arXiv2026

When Treatments Spill Over: A Neural Framework for Spatiotemporal Causal Effect Estimation

Real-world public health policies are often enacted by local jurisdictions, but their effects can ripple through connected populations. For causal evaluation, this creates a difficult spatiotemporal problem: intermittent policy updates create path-level treatments, and because spillover depends on neighboring treatment paths along with the focal-unit's own history, controlling time-varying confounding requires high-dimensional, multimodal information from both sources. To address this challenge, we develop a neural spatiotemporal spillover framework that extends potential-outcome causal analysis to networked policy-systems with unit-level effects and structured interference. The framework indexes each unit's potential outcomes through own-treatment and neighborhood spillover-exposure paths, yielding direct, spillover, and total effects. We introduce an LSTM-based representation-learning encoder that integrates each unit's own and neighborhood histories into compact sequential summaries for downstream modeling and causal estimation. These summaries feed a self-normalized kernel inverse-probability-weighting estimator for continuous spillover exposure. Beyond theoretical guarantees, simulations show improved performance over outcome-regression baselines; in a practically motivated county-level COVID-19 policy study, the framework finds both business-and-economic restrictions and education-and-childcare policies associated with lower subsequent case burden, with larger estimates for business-and-economic restrictions. Overall, it provides a scalable, policy-interpretable framework for evaluating unit-level effects in high-dimensional, multimodal spatiotemporal systems when interventions spill over across space and time.

stat.ME↗

arXiv2026

Bench2Robust: Scenario-Controlled Recovery for Tool-Using Agents

Tool-using LLM agents are typically developed and post-trained in controlled environments with reliable tools, but deployment exposes them to tool failures, API and schema drift, and third-party services outside the developer's control. We introduce Bench2Robust, a framework for scenario-controlled solvability that augments existing benchmarks with episodes requiring retrying the original path, switching to a fallback-equivalent tool, or recognizing that no viable path remains. By controlling recoverability, Bench2Robust separates recovery-policy quality from randomness in whether failures persist or resolve. Across 7 models from 4 families and 10 task/domain slices spanning customer-service and function-calling benchmarks, 69 of 70 model-subset pairs degrade under tool failures, by up to 46.7 percentage points. We propose Bayesian Tool Memory (BTM) for structured runtime recovery context and use reinforcement learning (RL) for learned recovery behavior. On held-out Retail tasks using the benchmark's original tool set, BTM improves robustness by 16.8 percentage points without retraining, while RL retains a 6.3-point gain without inference-time BTM. Combining RL with BTM yields a 20.7-point improvement over the base model while preserving clean performance. These results show that robust tool use benefits from combining structured runtime recovery context with learned recovery behavior.

cs.AI↗

arXiv2026

The Path Partition Conjecture and 3-(quasi-)Transitive Digraphs

A digraph $D = (V,E)$ is 3-transitive (respectively, 3-quasi-transitive) if whenever $(u,v,w,z)$ is a directed path of length 3, $(u,z) \in E$ (respectively, $(u,z) \in E$ or $(z,u) \in E$). These digraphs extend the concepts of transitive and quasi-transitive digraphs, respectively. For 3-transitive and 3-quasi-transitive digraphs, only the strong case has been characterized. In this paper, we introduce the notion of semi-acyclic digraphs and use it to describe completely the structure of 3-transitive and 3-quasi-transitive digraphs. As an application of this type of structure description, we prove that the path partition conjecture holds for the class of extended semi-acyclic digraphs, which generalizes two previously known results on the conjecture.

math.CO↗

arXiv2026

Three-Phase Unbalance Mitigation via DSO-FRA Coordination: A GNB-Based Chance-Constrained Model Considering PV Uncertainty

Although flexible resource aggregators (FRAs), represented by electric vehicle aggregators (EVAs) and load aggregators (LAs), possess inherent flexibility to mitigate three-phase unbalance, the key challenge lies in how to effectively incentivize them to actively participate in unbalance mitigation. To address this, this paper proposes a generalized Nash bargaining (GNB)-based chance-constrained coordinated operation model for the distribution system operator (DSO) and FRAs. In this framework, the DSO and FRAs cooperate to mitigate three-phase unbalance, where FRAs provide flexibility to reduce the DSO's unbalance mitigation costs, and the DSO offers economic incentives to FRAs in return. Bargaining power is tailored according to each participant's contribution to unbalance mitigation to ensure fair profit allocation. Moreover, a scenario-based chance-constrained formulation is integrated to handle photovoltaic (PV) output uncertainties. To solve the model, a distributed proximal decomposition algorithm (PDA) is introduced for the independent model, while the coordinated model is decomposed into a social welfare maximization subproblem and a payment bargaining subproblem, with an improved bilinear Benders decomposition algorithm developed to solve the former. Numerical results validate the effectiveness of the proposed method in mitigating unbalance, ensuring fair profit distribution, and enhancing robustness against uncertainties.

math.OC↗

arXiv2026

Agentic Tool-Augmented Reasoning for Explainable Image Forgery Detection

Conventional image forgery detection methods produce binary scores or pixel-level masks without interpretable evidence, while recent multimodal large language model (MLLM)-based approaches generate post-hoc explanations of predetermined classification results rather than reasoning from evidence. Inspired by the forensic workflow of human judicial experts, we propose Agentic Tool-Augmented Reasoning (ATAR), a framework integrating 22 specialized forensic tools across seven complementary domains to autonomously detect, localize, and explain image forgeries through multi-turn reasoning. A Dual-Stream Forensic Reasoning paradigm combines a high-level semantic anomaly path, which magnifies suspicious regions for fine-grained inspection, with a low-level forgery artifact path, which invokes forensic tools to extract objective evidence. We further introduce Forensics Curriculum Learning: during General Experience SFT, an automated teacher-student mentoring pipeline synthesizes multi-turn tool-usage reasoning trajectories; during Forensic Scene RL, a Tool Prior Curriculum guides early tool exploration and progressively transfers control to the agent, while a Structured Evidence Reward provides fine-grained process-level supervision. Experiments across IMDL, Deepfake detection, DMDL, and AIGC detection show that ATAR achieves 78.5% average image-level F1 on six zero-shot IMDL benchmarks, surpassing the strongest MLLM baseline by 11.8 percentage points, and remains competitive with specialized detectors on other tasks while producing substantially more faithful and grounded explanations.

cs.CV↗

arXiv2026

SpanUQ: Span-Level Uncertainty Quantification for Large Language Model Generation

Uncertainty estimation is essential not only for the trustworthy deployment of large language models (LLMs) but also as a foundation for self-refinement in LLM generation. However, existing approaches operate at suboptimal granularities: token-level scores lack semantic coherence, while sequence-level scores fail to localize errors. We formalize Span-Level Uncertainty Estimation (SLUE), a new task that targets the natural granularity for uncertainty: semantically coherent text spans, each conveying a single assessable unit of meaning. To address this task, we introduce SPANUQ, a lightweight (25M parameter) probe that distills the uncertainty knowledge from expensive multi-sample inference into a single forward pass over LLM hidden states. SPANUQ employs a DETR-style span decoder to simultaneously detect spans and estimate their uncertainty via a Mixture of Beta distribution, trained with a principled combination of Beta NLL regression and contrastive ranking objectives. We construct SPANUQ-BENCH, the first span-level uncertainty benchmark comprising 20K prompts, 293K annotated spans, and continuous soft labels derived from multi-sample claim verification. Experiments on five LLM backbones show that SPANUQ consistently achieves the best span-level uncertainty quality , outperforming the strongest probe baseline and all sampling-based methods while being 10~20x faster. Its DETR-based span detector attains 0.910 F1, surpassing the best heuristic by 39.4%, enabling precise error localization that sequence-level methods cannot provide. The same architecture ports to five LLMs spanning two model families, with one probe trained per backbone, and we additionally observe that sequence-level uncertainty is partially decomposable, suggesting that span-level estimation subsumes sequence-level as a special case. The project page is available damon-demon.github.io/SpanUQ.

cs.CL↗

arXiv2026

PolyOCR-Venus: Unified OCR Foundation Models for Text-Centric Visual Intelligence

Optical Character Recognition (OCR) is evolving from plain-text transcription toward general visual intelligence, requiring models to recognize, localize, and reason over textual information in complex visual environments. However, existing OCR systems often excel at only some tasks and struggle to balance recognition, parsing, and reasoning across scenarios. In this report, we present PolyOCR, a family of unified OCR foundation models of varying scales. PolyOCR combines a shared instruction-following framework with a large-scale data engine that converts heterogeneous visual resources into quality-verified OCR supervision. We introduce Competence-Guided Policy Optimization, which combines verifier-based Group Relative Policy Optimization with on-policy distillation through sample-wise routing based on teacher reliability and the teacher--student competence gap. We also introduce OCRBench v2.1, our revision of OCRBench v2 with manually verified annotation corrections and task-aligned scoring metrics. Extensive experiments across OCRBench v2.1, CC-OCR, in-house KIE Benchmark, OmniDocBench v1.6 and MDPBench demonstrate that PolyOCR achieves state-of-the-art or highly competitive performance.

cs.CV↗

arXiv2026

In-Context Learning for Robots: Methods and Applications

General-purpose robots must infer what a new task requires and translate that understanding into appropriate physical action. In-context learning (ICL) for robots supports this process by using demonstrations and interaction to direct existing competence with neural parameters held fixed during deployment. We organize this literature review around the interfaces connecting contextual evidence to execution, distinguishing four families: context-conditioned policies, geometric demonstration transfer, world-model-based control, and skill- and agent-based execution. Comparing these interfaces clarifies their transfer assumptions and the roles of training, correspondence, and memory in making context useful. Across manipulation and navigation, we examine how these mechanisms preserve taught requirements as objects, environments, and execution conditions change. This analysis links method design to evaluation practices that distinguish responsiveness to teaching, physical transfer, and benefits from retained experience. The resulting agenda connects compositional task acquisition and faithful transfer with physical recursive self-improvement, in which experience improves the ability to learn subsequent tasks.

cs.RO↗

arXiv2026

The Erdős--Chvátal Simplex Conjecture

We prove the Erdős--Chvátal simplex conjecture across its entire parameter range. If $2\le d<k$ and $(d+1)k\le dv$, then every $d$-simplex-free family of $k$-subsets of $[v]$ has at most $\binom{v-1}{k-1}$ members, with equality only for a full star.

math.CO↗