arXiv subjects

Xiaoyang Wang

Publications and source records attributed to Xiaoyang Wang.

At least 19 recordsLinked to original sources

arXiv2026

Ctrl-G: Controlling Structural and Interaction Exposure in Knowledge Graph Augmented LLM Reasoning

Knowledge graphs (KGs) provide structured evidence for Web question answering with large language models (LLMs), but sending retrieved evidence to a remote LLM can expose sensitive graph information. Existing privacy treatments use entity anonymization and relation-centric abstraction to preserve reasoning cues, but transmitted topology remains visible and disclosure accumulates across repeated calls. Distinctive topology can support re-identification of masked entities (structural exposure), while successive calls accumulate additional paths and exploration traces into a broader view of the KG (interaction exposure). Simply sending less evidence is insufficient because restricted, stepwise views may omit the bridge paths required for multi-hop reasoning. To address this, we propose Ctrl-G, to the best of our knowledge the first KG-RAG framework to jointly control structural and interaction exposure to a remote LLM while supporting accurate multi-hop reasoning. To control structural exposure, Ctrl-G locally constructs query-conditioned sanitized views through structural aggregation and pruning, limiting the graph evidence transmitted to a remote LLM. A hierarchical controller couples remote planning with local connected-path exploration across topic entities, using question indicators and retrieved experience to prune candidates while grounding and verifying retained evidence against the original KG. To control interaction exposure, local experience memory reuses abstract, verified reasoning traces to guide exploration modes and depths and gate remote LLM calls, reducing repeated disclosure across questions. Ctrl-G achieves the highest accuracy among the compared methods on five of six benchmarks, exceeding the strongest baseline by up to 13.8 points across model configurations.

cs.CL↗

arXiv2026

BlindStep: A Secure One-Step Kernel for Privacy-Preserving Collaborative Reasoning

A recurring step in two-party collaborative workflows combines a private candidate frontier with a partner's compatible evidence, scores the resulting candidates, and retains only a small frontier. Direct execution can expose intermediate candidates, local scores, or unauthorized evidence provenance. Existing MPC platforms and PSI protocols provide useful building blocks, but they do not define a unified post-alignment boundary for bounded-slot aggregation, fail-closed overflow handling, retained-frontier selection, and downstream disclosure. We present BlindStep, a fixed-shape MPC-online kernel for one such post-alignment step between two semi-honest parties. Given a frontend realizing the bounded-slot contract and an application enforcing the reveal policy $ρ$, BlindStep provides a per-invocation simulation-based guarantee in the $F_{\mathrm{align}}$-hybrid model: the kernel leaks only public setup parameters, fixed execution-shape metadata, a global overflow bit, and $ρ$-authorized Layer-2 disclosures. The guarantee excludes malicious security and multi-invocation composition, and treats frontend leakage and offline preprocessing separately. Within this kernel boundary, we implement six semi2k realizations in MP-SPDZ, crossing two sharing styles and three top-K strategies: H-variants compare over arithmetic shares, while B-variants convert packed ranking values to Boolean shares before selection. Under LAN, B-Queue is 6-7 times faster and uses about 40 times less communication than H-Queue at N = 256 and 2048. Against H-Sort on the same backend, B-Queue achieves speedups of 21 and 30 times and communication reductions of 204 and 290 times at N = 2048 and 8192, respectively. Under the evaluated N = 256, 100 ms RTT WAN profile, H-Sort becomes fastest, showing that round complexity can dominate backend choice for the measured MPC-online kernel.

cs.CR↗

arXiv2026

Disparity Laundering: Consensus Without Fairness in Clinical Multi-Agent Systems

Clinical large language models are increasingly fielded as multi-agent teams, on the assumption that deliberation averages out individual error. It does not average out bias. Across a full factorial of five model families and four aggregation topologies on 120 real MIMIC-IV emergency cases under counterfactual demographic swaps, no configuration neutralizes the disparity: all 20 cells commit the direction of their agents' independent answers, giving Black patients more acute triage. Every model has a topology that amplifies its baseline gap, but which one varies by model, so no design is reliably safe or reliably amplifying: the disparity is relocated, never removed. Late-stage fixes fail: fairness instructions and demographic masking leave the gap untouched, and per-agent self-debiasing widens it, because the bias is already encoded in the agents' ordinary clinical numbers. We call this disparity laundering; the audits now applied to clinical AI, per model, per output, per version, can certify the parts but never the committed decision. Interventions close the gap exactly when they make counterfactual versions visible inside the system and pool them, through deliberation, into the committed decision. A diverse-perspective ensemble does both while keeping the team, with no retraining or architectural change: the disparity falls by 67 to 94% where all five perspectives pool at every round, and the one topology that never pools them is the one it cannot reliably repair. Medical multi-agent systems do not filter demographic bias; they launder it into plausible decisions no single-version audit can catch, and the only defense is to make the bias visible.

cs.AI↗

arXiv2026

How AI Load Swings Reduce Datacenter Power Conversion Efficiency

AI training and inference workloads make GPU power swing repeatedly between compute and communication phases. As a result, the cascaded converters in a datacenter rack keep moving between operating points with different efficiencies. This paper uses an energy based, cycle averaged method to quantify that effect along a 240 V AC to 400 V DC to 51 V DC rack power path, using measured efficiency curves of commercially available GaN based PFC and LLC converters. For a 1 Hz, 50% duty-cycle load swinging between 5 kW and 510 W, conversion efficiency falls to 95.97%, compared with 96.82% at the equivalent constant average load, a loss of 0.88%. When 51/48V to 1V point of load conversion to the chip rail is included, it is shown that the loss grows to 1.66% (89.45% vs. 90.96%). These results show that efficiency rated at average load overstates real world performance, and that the penalty grows with each load dependent conversion stage between the grid and the GPU.

cs.AR↗

arXiv2026

Do Agents Need a New Kind of Paper? SPRA: Structured Paper Representation for Agents

Scientific agents must connect research problems, methods, and findings dispersed through narrative articles. We introduce SPRA (Structured Paper Representation for Agents), a paper-level representation of persistent research objects and typed relations. Its empirical-AI profile organizes Problems, Contributions, Components, ExperimentSetups, Measures, and Findings for querying, traversal, and reuse. Selected paper-level citation relations record dependencies, component use, and comparisons with prior work; source locations support verification. We instantiate SPRA across more than one million AI papers, with 19.95 million research units and 24.07 million internal relations. Evaluations access SPRA through Schema-RAG with task evidence fixed before inference. Across seven subfields, SPRA improves source-conditioned scientific QA by 0.381/5 over passage retrieval while remaining close to Full Compact (0.078/5 lower on average). In directed cross-paper reasoning, qualification improves by 0.184/5 over passage retrieval; Top-5 synthesis shows mixed quality effects. Across all paired RQ2 tasks, including Top-5, input+output tokens decrease by 28.5% relative to passage retrieval and 95.3% relative to Full Compact. In idea generation, a 0.067-point lower Overall score than Full Compact accompanies 93.7% fewer input+output tokens. Together, these results support SPRA as a usable representation of scientific content and a foundation for research workflows that connect, inspect, and reuse paper-level research objects. The SPRA resource is accessible through the Sciverse scholarly-data platform.

cs.AI↗

arXiv2026

Bandit Multiclass PAC Learning: Structural Baselines and a Confidence Direct Sum

How much does high confidence cost when a multiclass learner sees only whether its prediction is correct? We separate two behaviors in realizable bandit PAC learning. An anchored dimension gives a general lower bound that is attained, up to universal constants, by a family of product-block classes. In contrast, for $n\ge16$ divisible by four, an affine multiplexer with bandit DS dimension $2n-1$ has sample complexity $Θ((n\min\{n,\ln(1/δ)\}+\ln(1/δ))/\varepsilon)$ in the standard accuracy and confidence regime. Its lower bound applies to adaptive randomized learners with improper outputs. The obstruction is that evidence collected against one candidate group cannot be reused against the others. The better of two algorithms matches this cost. One finds the correct group through a positive response. The other exhausts the rank of successive groups. The resulting confidence profile rules out a universal additive-confidence bound linear in the bandit DS dimension. Moreover, classes with identical anchored dimension, bandit DS dimension, and pointwise label width can have polynomially different sample complexities. We also sharpen the confidence accounting in ListCascade and show how one positive response replaces the bandit DS dimension by the anchored dimension in a general upper bound.

cs.LG↗

arXiv2026

An Agnostic Sample Compression Scheme for Squared Loss of Near-Linear Size in the Fat-Shattering Dimension

We study the squared-loss sample compression question posed by Attias, Hanneke, Kontorovich, and Sadigurschi in their ICML 2024 paper. For every nonempty class $\mathcal{F}\subseteq[0,1]^{\mathcal{X}}$ and accuracy $0<α\le1$ with $d=\operatorname{fat}(\mathcal{F},c'α)<\infty$, where $c'>0$ is universal, we prove the existence of a deterministic agnostic scheme of total size $O(d\log^3(2/α))$. Its code contains original labeled observations and auxiliary bits, and its reconstruction has empirical loss at most $α$ above the infimum over $\mathcal{F}$ on every finite sample. This establishes the polylogarithmic formulation in Section 5 of arXiv:1810.01864v2. The construction approximates a near-optimal internal target by a fixed-step boosting ensemble and takes its equal-weight median. Average loss permits an accuracy-dependent number of rounds, so the entire ensemble can be retained without dual-dimension sparsification. Quantized target constraints and round-membership bits determine the reconstructed function exactly. We also prove that, when retained observations must come from the sample, a fixed class of fat-shattering dimension one requires total size $Ω(\log(1/α))$. With observations and auxiliary bits charged together, this rules out the proceedings version's log-free formulation under its original-example restriction. The framework extends to average power loss at scale $α/p$.

cs.LG↗

arXiv2026

Learning Quantum Operator Dynamics from Short-Time Data

Real-time dynamics of quantum observables provide direct access to excitation spectra and correlation functions in quantum many-body systems, but currently available quantum devices are limited to short evolution times due to decoherence. We propose a neural ordinary differential equation (Neural ODE) framework with a physics-driven neural network to reconstruct long-time operator dynamics from short-time quantum data. By expanding observables in the Pauli basis and exploiting locality and symmetry constraints, the operator evolution is reduced to a tractable set of coefficients whose dynamics are learned from the measurement data. Applied to the Ising spin chain with mixed fields, the operator learning method accurately extrapolates its long-time correlation function and resolves its excitation spectrum. We validate the framework by collecting short-time data from IBM's superconducting quantum processor and successfully reconstruct the Fourier spectrum of an $N=8$ Ising spin chain with 5 times enhanced precision. Our results demonstrate a scalable and data-efficient strategy for extracting dynamical and spectral information from practical quantum hardware.

quant-ph↗

arXiv2026

ANTE: Anticipate, Then Execute -- Learning Internal World Models for Coding Agents from Verified Beliefs

Rapid advances in large language models have made software engineering agents capable of tackling complex coding tasks. Within a containerized environment, a coding agent receives tool responses or execution results, and determines its later actions by those earlier results. However, most existing coding agents never look ahead: their policies do not anticipate the consequences of actions, and the terminal reward barely attributes a failure to certain earlier steps, so an ill-judged edit can corrupt the repository state irreversibly before any signal arrives. Recent code world models aim to address these challenges, but they are either trained and then frozen before reinforcement learning (RL), or reduced to an auxiliary next-token loss on the environment's outputs. We present ANTE (Anticipate, Then Execute), an agentic RL framework that verifies the policy's pre-action predictions by real execution and turns them into per-step credit for its actions and trust-or-abstain decisions. With an internal belief stated before each action, Belief-Anchored Policy Optimization (BAPO) turns the settled predictions into per-step advantages by anchoring credit on the policy's own forecasts. To turn the trained forecasts into better actions at inference, thereby facilitating effective test-time scaling, ANTE-TTO (test-time optimization) selects among candidate actions without an external scorer. ANTE raises Qwen3-32B from 25.0% to 57.2% pass@1 on SWE-bench Verified, 6.2% above a matched GRPO baseline trained without the belief state; its world model's accuracy climbs from 44% to 69% during RL, and the learned foresight capability transfers across harnesses and out-of-distribution domains. In summary, we aim to build a code world model that knows what has happened, what holds now, and what the next action will achieve.

cs.LG↗

arXiv2026

GPF-Net: Gated Progressive Fusion Learning for Polyp Re-Identification

Colonoscopic Polyp Re-Identification (ReID) aims to match the same polyp across a large gallery of images captured from different viewpoints and with different cameras, playing a critical role in computer-aided diagnosis for the prevention and treatment of colorectal cancer. However, the coarse granularity of high-level features often limits performance on small polyps, where fine-grained details are essential for accurate matching. To address this challenge, we propose a novel multimodal feature fusion architecture, termed the Gated Progressive Fusion Network, which selectively integrates features from multiple levels through fully connected gating mechanisms. Building on this framework, we introduce a gated progressive fusion strategy that enables layer-wise refinement of semantic information, facilitating multi-level feature interactions to enhance both generalization ability and robustness. Extensive experiments on standard benchmarks demonstrate the advantages of the multimodal setting over state-of-the-art unimodal ReID models, particularly when combined with the proposed fusion strategy tailored for general-purpose scenarios.

cs.CV↗

arXiv2026

Knowledge boundary probing and demand-guided intervention for LLM-based power system code generation

Large language models (LLMs) can turn grid-analysis requests into executable programs for power-system simulation, but utilities and research laboratories often require on-premise deployment. In this setting, first-pass failures frequently arise at an API-knowledge boundary, through hallucinated functions, misused parameters, and mishandled result tables. We present PowerCodeBench, a parameterised benchmark generator released as a frozen 2,000-task suite for pandapower, and a deployment-time workflow that requires no weight updates. Documentation-driven L0-L3 probes produce per-model API profiles for diagnosis, model comparison, documentation allocation, and backend calibration. A query-side demand estimator selects layered API evidence before generation, while execution feedback routes targeted repair. Across ten open-weight LLMs (1.5B-480B) and four mid-tier APIs, the validation-enabled workflow raises scalar-match accuracy by 32-56 percentage points after up to three repair rounds relative to an unassisted first pass, for every model of at least 7B and every API. Open-weight models in the 70B-120B range reach the four-vendor mid-tier accuracy range under matched no-tool conditions. Selective injection approaches the full-layer reference using 41% of its prompt tokens. Among model-item pairs passing numerical checks under both workflows, engineering review confirms the requested analysis in 88% of full-workflow outputs versus 66% under plain repair. Round-0 pilots on OpenDSS and PyPSA motivate staged onboarding from broad retrieval at cold start to calibrated selective injection. Measured throughput, latency, energy, and allocated GPU memory establish a practical on-premise serving envelope.

cs.SE↗

arXiv2026

Unanimously Wrong: Certified Abstention from How Medical LLM Consensus Forms

In clinical practice, agreement among independent experts is treated as evidence of reliability, and multi-round consensus has become a core mechanism of agentic medical question-answering systems. When such a system must decide whether to trust its own answer, the prevailing signal is again agreement, now among the sampled answers. But agreement is a fragile proxy for correctness. A system can be unanimously wrong, returning the same incorrect answer on every sample, and on these questions agreement-based signals carry no information. The cause is that these signals read only the final state of the consensus and discard how it was reached. Agreement that was reached by resolving disagreement with evidence looks identical, at the end, to agreement that was present from the first sample because every sample shares one misconception. ProbeGuard is a certified abstention framework that bases the abstention decision on how the consensus formed. Process features trace agreement trajectories, minority persistence, and retrieval saturation. For unanimous votes, rationale semantic entropy checks whether the reasons behind the vote cohere, and an active probe retrieves counter-evidence and measures whether the consensus survives. A stratified Learn-then-Test calibration then converts these scores into a distribution-free bound on selective risk. We evaluate ProbeGuard on three medical QA benchmarks and a hard-frontier reference, with a published multi-round agentic RAG substrate, against six abstention baselines. On MedQA, 13.4% of unanimous votes are wrong, and no agreement-based signal can flag them. Process signals raise the discrimination of correct from incorrect consensus from chance to 0.696 AUROC. The certified rule answers six in ten unanimous-layer questions at an observed selective risk of 9.0%, and nine in ten once in-domain calibration data accumulate.

cs.AI↗

arXiv2026

Do Higher-Order Models Win for Higher-Order Reasons? Rethinking Performance Gains in Hypergraph Learning

Higher-order models (e.g., hypergraph neural networks) often outperform lower-order baselines on hypergraph learning benchmarks, and their advantages are commonly attributed to their ability to exploit higher-order information. However, better performance alone does not establish this explanation. We therefore ask: Do higher-order models win for higher-order reasons? To investigate this question, we introduce a controlled performance-attribution framework that perturbs higher-order information while preserving the lower-order, i.e., pairwise, information. Across 25 commonly used hypergraph learning benchmarks spanning three tasks, we frequently observe an intriguing pattern: higher-order models originally outperform lower-order baselines, yet retain most of their advantage after perturbation. This suggests that much of the observed advantage remains achievable without the higher-order information. We then investigate potential lower-order explanations for these remaining gaps. We find that simple additions to a lower-order baseline, e.g., richer pairwise weighting, more steps of pairwise feature propagation, and normalization, reduce the remaining performance gaps, supporting lower-order explanations for part of the observed advantage. Our analysis calls for the hypergraph learning community to rethink performance attribution by distinguishing performance gains from their explanations, adopt stronger lower-order baselines, and use suitable benchmarks that better test the value of higher-order information.

cs.LG↗

arXiv2026

Streaming the partial-transpose moment hierarchy with order-independent quantum memory

Partial transposition reveals mixed-state entanglement through the spectrum of $ρ_{AB}^{T_B}$, but this operator is not generally a physical state. Its power moments are accessible through collective permutations, whose coherent width grows with moment order. Here we introduce a streaming protocol that processes fresh copies sequentially while retaining one copy and a reusable ancilla. By realizing opposite cyclic permutations on the two subsystems, each measurement record yields unbiased estimators of all moments $p_2,\ldots,p_K$ through successive outcome products. For an $n$-qubit state, the protocol uses $2n+1$ active qubits independent of $K$ and $O(K\log K/ε_{\rm mom}^2)$ copies, matching $Ω(K/ε_{\rm mom}^2)$ lower bounds up to a logarithmic factor, including a PT-specific isospectral construction. Higher-order moments strengthen thermal-state entanglement certification toward the exact PPT threshold. Simulations cover pure and mixed states, and a cloud quantum processor implements a three-qubit instance.

quant-ph↗

arXiv2026

Uniform Observability of High-Order Fully Discrete Multidimensional Wave Equations: Exponential Loss and Recovery

Stable discrete approximation of wave controls requires observability estimates that remain uniform under mesh refinement. Such uniformity is known to fail for many low-order discretizations, but the quantitative understanding of observability loss for arbitrary high-order approximations is largely open, owing to their multiple spectral branches and the absence of explicit dispersion relations. To address this gap, we establish a quantitative theory of observability loss and recovery for fully discrete $Q^k$ local discontinuous Galerkin approximations of the multidimensional wave equation, valid for any fixed polynomial degree $k$. Our analysis of the $(k+1)^d$-branch discrete spectrum identifies a unique physical branch that recovers the continuous propagation speed at low frequencies. Under a strict stability margin on this branch, its group velocity vanishes at high-frequency symmetry points. To quantify the resulting loss of uniform observability, we prove that the observability constant satisfies an exponential lower bound of the form $\exp(Ch^{-(1-\varepsilon)})$ for any $\varepsilon\in(0,1)$ by constructing Gevrey-localized wave packets. We then provide a recovery strategy that restores uniform observability for any fixed polynomial degree $k$, with a sufficient observation time controlled by the minimum retained physical group velocity. Finally, our numerical results suggest that higher-order approximations retain a broader genuine physical-frequency band at a prescribed positive group-speed threshold, allowing less restrictive filtering in wave-control computation and better preserving wave dynamics.

math.OC↗

arXiv2026

From Trajectories to Grounded Preferences: Process Preference Synthesis via Interaction Element Graphs for Web PRMs

Comparative Process Reward Models (PRMs) provide critical step-level guidance for autonomous web agents by evaluating state-conditioned preferences between candidate actions. However, existing preference training data synthesized via multi-policy sampling suffers from a severe scarcity of Grounded Minimal Contrastive Pairs (GMCPs)-where competing candidates target genuine on-page elements with identical action types. In representative baselines preference data (namely, WebArbiter), GMCPs account for merely 24.19%, biasing PRMs during training to rely on shallow shortcuts (such as element hallucinations and action type mismatches) rather than acquiring genuine contextual decision semantics. To address these challenges, we propose SURFPRM, a graph-guided process preference synthesis framework for comparative Web PRMs. SURFPRM structures web demonstrations into a persistent Interaction Element Graph that acts as an environment-grounded negative action proposal mechanism, systematically synthesizing contrastive negative actions across spatial, temporal, and spatiotemporal confusion axes. This elevates the GMCP proportion from 24.19% to 74.60%, producing the curated SURFPRM-DATA dataset. Across six open-source backbones (3B to 9B parameters), PRMs trained on SURFPRM-DATA outperform baseline-trained models on average on WEBPRMBENCH and rival leading proprietary LLMs. In downstream reward-guided trajectory search on WEBARENA-LITE, SURFPRM provides step-level guidance for both GPT-4o (+14.21%) and GPT-4o-mini (+12.83%) policies, yielding substantial improvements in complex web task success rates.

cs.AI↗

arXiv2026

Efficient Quantum Simulations of Yang-Mills theory with Maximal-tree Gauge

We develop a quantum algorithmic framework for the efficient simulation of Yang--Mills theories, including the $\mathrm{SU}(3)$ gauge theory in Quantum Chromodynamics (QCD). The framework uses maximal-tree gauge in terms of gauge field variables that removes all local gauge redundancies. In the resulting gauge-fixed formulation and digitization in the field-amplitude basis, we show that Hamiltonian time evolution admits an efficient implementation based on quantum singular value transformation (QSVT). We derive upper bounds on the total number of qubits and gate complexity, finding polynomial scaling with the inverse simulation precision $1/\varepsilon_s$, lattice volume $\mathcal{V}$, gauge coupling $g$, and target energy scale $E$. Our results provide a rigorous complexity-theoretic demonstration that non-Abelian Yang--Mills theories can be simulated efficiently on quantum computers, paving the way toward first-principles quantum simulations of non-perturbative QCD dynamics.

hep-lat↗

arXiv2026

Action Forcing: Training World Models on Unsupervised Video by Recovering Underlying Egomotion Bases

Synchronised action annotations are needed to train controllable world models and these datasets remain elusive. Existing approaches make use of instrumented platforms with calibrated sensors, costly manual annotation, or latent-action models which lack grounding. We instead turn ordinary unlabelled video into action-supervised training data by recovering (without training) a data-derived egomotion basis. We track pixel displacements across frames and exploit the recurring coherent structure induced by egomotion to obtain grounded control signals directly. Using a method as simple as principal components analysis perform this, we find that the leading components provide signed, scalable, and composable throttle--yaw controls, although the method can recover only motion axes represented in the data. To prevent a high-capacity video DiT from exploiting pixel-level supervision, an online latent critic distils a frozen decoder--tracker--PCA (Principal Components Analysis) teacher without backpropagating through the decoder or tracker. Finally we critique the use of video generation metrics to evaluate WMs and introduce an example of an alternative, reference-free evaluation method. We measure \textit{controllability}, \textit{plausibility}, \textit{conjuring} (creating objects out of thin air) and \textit{geometric integrity}, revealing failures that conventional video metrics miss. We show that most baselines follow familiar action directions but struggle to reverse or remain stationary. Our model handles both while retaining compositional control and generation quality. Despite backwards actions being less than $1\%$ of our training data, we find that the model learns to reverse, scale its response linearly, and compose throttle with steering, all simply by learning through a grounded action space.

cs.CV↗