arXiv subjects

Xiaodong Wang

Publications and source records attributed to Xiaodong Wang.

At least 37 records · Page 2Linked to original sources

arXiv2026

SPID-Chain: Verifiable Polar-Coded State Validation for Cross-Chain DAG Settlement

Cross-chain settlement must preserve safety across heterogeneous ledgers while tolerating delayed computation, Byzantine participants, and adversarial transaction issuance. This paper presents SPID-Chain, an adapter-compatible settlement architecture for escrow-backed fungible transfers across programmable blockchains. SPID-Chain maintains settlement state through persistent Polar-coded fragments, validates candidate state transitions using hidden linear verification checks, and records certified transfers in a weighted directed acyclic graph (DAG). The design separates native-chain finality from cross-chain settlement: source-chain finality establishes an immutable reservation, whereas weighted DAG confirmation determines when the corresponding destination credit becomes executable. We derive an exact recovery-time distribution for heterogeneous coded workers, a verification-soundness bound for Byzantine responses, and an exact weighted-quorum condition for conflicting-block safety. These components are coupled in a cross-layer stability theorem showing how the coded-validation completion probability determines the effective honest issuance rate and, consequently, the stable adversarial-load region of the settlement DAG. We further establish an end-to-end settlement guarantee covering balance non-negativity, asset conservation, conflict exclusion, replay protection, coded-state consistency, and finite expected lock-to-release latency under the stated liveness conditions. Prototype-assisted simulations indicate that coded validation reduces sensitivity to stragglers, improves validation and confirmation throughput under heterogeneous delays, and produces the predicted transition between stable and unstable DAG operation. The resulting framework provides a verifiable and analytically grounded settlement layer without modifying the native consensus protocol of participating chains.

cs.DC↗

arXiv2026

CMSL: Constructive Multi-Sequence Learning for Recommendation Systems

Sequence learning has emerged as the promising paradigm in recommendation systems, surpassing traditional Deep Learning Recommendation Models (DLRM) by capturing the temporal nuances of user behavior. However, current state-of-the-art architectures operate under a limiting analogy: they treat user history as a monolithic chronological sequence like a sentence in a Large Language Model (LLM). We observe a fundamental divergence between natural language and recommendation data: unlike the linear, logical flow of text, user history is inherently multi-faceted. A user's journey is a fragmented reflection of diverse interests, resulting in much weaker coherence between items than is found in LLM training data. This lack of structural unity leads to context pollution. In single-sequence modeling, unrelated behaviors compete for the same attention budget. This "noisy" signal dilutes the model's focus, effectively capping its ability to discern high-intent patterns from background activity. To address this, we propose Constructive Multi-Sequence Learning (CMSL), a paradigm shift from passive sequence ingestion to active "context engineering" that constructs multiple coherent sequences in latent space. CMSL leverages a learnable Sequence Construction Module to disentangle user history into "pure" thematic strands, followed by a linear attention mechanism to efficiently model these strands at scale. CMSL has been deployed across ranking and retrieval tasks and across four major surfaces at Meta.

cs.IR↗

arXiv2026

Secure Decentralized Federated Learning via Gossip and Virtual Voting

Decentralized federated learning (DFL) removes the central server by letting nodes exchange model updates through peer-to-peer gossip, but existing gossip-based methods often lack provenance finality and resilience to Byzantine or lazy participants. Ledger-assisted federated learning (FL) improves auditability, yet blockchains, shards, or settlement committees can reintroduce global coordination costs that conflict with DFL locality. This paper proposes \emph{gspDAG-FL}, a secure DFL framework that derives consensus from the same gossip history used to disseminate models. Nodes exchange model payloads only with neighbors, while full nodes collect event certificates and receiver-endorsed accepted gossip proofs, reconstruct a compact Topology directed acyclic graph (DAG), and run Hashgraph-style virtual voting followed by compact full-node certificates. Finality is over unique model-origin tuples, not identical local parameter states. To improve resilience, gspDAG-FL combines payload validation, accepted-proof validation, and private semantic audit before aggregation. We formalize the adversarial setting, prove safety and conditional liveness of the control plane, and give a convergence guarantee for certified perturbed gossip under time-varying effective mixing. Experiments on MNIST classification and Penn Treebank language modeling, using fair held-out validation/audit data and networks up to $N=100$, show that gspDAG-FL achieves learning quality close to validation-based ledger FL while reducing coordination bottlenecks, improving throughput, and maintaining high invalid-origin detection under mixed Byzantine and lazy participation.

cs.LG↗

arXiv2026

Keyless Covert Communication Over Quantum MACs with General Message Sets

We study covert classical communication over quantum multiple-access channels (MACs) with general message sets. Specifically, we consider a fully quantum MAC with arbitrary message sets and an arbitrary number of transmitters. We demonstrate the feasibility of achieving a positive covert rate over this channel and establish general one-shot and asymptotic achievable rate regions. For classical-quantum MACs with general message sets, we establish the covert capacity, when the transmitters are restricted to deterministic encoding. Our result recovers, as a special case, known results for classical communication over classical MACs with general message sets, covert communication of a classical message over a classical channel with two transmitters, and classical communication over quantum MACs. We provide three examples of MACs to which our results can be applied, either directly or indirectly, to achieve positive covert rates. Specifically, we first study covert communication over a finite-dimensional MAC with a helper. We then analyze a classical Gaussian MAC with a helper and derive its covert capacity. Finally, we extend the analysis to a single-mode bosonic MAC with a helper and show that positive covert rates can also be achieved in this setting. To the best of our knowledge, this is the first work to achieve positive-rate covert communication over both classical and quantum MACs.

cs.IT↗

arXiv2026

Covert Communication Over a Quantum MAC with a Helper

We study covert classical communication over a quantum multiple-access channel (MAC) with a helper. Specifically, we consider three transmitters, where one transmitter helps the other two transmitters communicate covertly with a receiver. We demonstrate the feasibility of achieving a positive covert rate over this channel and establish an achievable rate region. Our result recovers as a special case known results for classical communication over classical MACs with a degraded message set, classical communication over quantum MACs, and classical communication over MACs with a helper. To the best of our knowledge, our result is the first to achieve covert communication with positive rates over both classical and quantum MACs.

cs.IT↗

arXiv2026

Ontology-Guided Evidence Path Inference for Multi-hop Knowledge Graph Question Answering

Knowledge graph question answering (KGQA) aims to answer natural-language questions by reasoning over structured facts. Existing multi-hop KGQA methods mainly rely on topic-centered expansion, which faces two key challenges: the search space rapidly grows with noisy mixed-type paths, and retrieved paths may fail to satisfy the semantic constraints of complex questions. To address these challenges, we propose OPI, an ontology-guided evidence path inference framework for multi-hop KGQA. OPI introduces a relation-centric ontology graph to capture the head-tail type constraints of relations, providing a compact interface for answer-side constraints. Based on this ontology graph, OPI first introduces a bidirectional retrieval mechanism by mapping the predicted answer type to compatible final-hop relations and combining topic-side prefix expansion with answer-side final-hop matching, thereby suppressing noisy mixed-type expansion. OPI further adopts an iterative refinement strategy to reassess retrieved paths and candidate answers under the question context, filtering type-compatible but question-irrelevant evidence for more reliable answer prediction. Experiments on WebQSP, CWQ, and MetaQA show that OPI substantially reduces the search space, improves Hit@1/F1 by 4.6/5.0 points on WebQSP and 8.9/3.3 points on CWQ over the strongest prior results, and achieves near-saturated Hit@1 on MetaQA with the retrieval module alone.

cs.AI↗

arXiv2026

SAM 3D: 3Dfy Anything in Images

We present SAM 3D, a generative model for visually grounded 3D object reconstruction, predicting geometry, texture, and layout from a single image. SAM 3D excels in natural images, where occlusion and scene clutter are common and visual recognition cues from context play a larger role. We achieve this with a human- and model-in-the-loop pipeline for annotating object shape, texture, and pose, providing visually grounded 3D reconstruction data at unprecedented scale. We learn from this data in a modern, multi-stage training framework that combines synthetic pretraining with real-world alignment, breaking the 3D "data barrier". We obtain significant gains over recent work, with at least a 5:1 win rate in human preference tests on real-world objects and scenes. We will release our code and model weights, an online demo, and a new challenging benchmark for in-the-wild 3D object reconstruction.

cs.CV↗

arXiv2026

OmniHalluc-L: Counterfactual Benchmarking and Modality-Perturbation Reliability Calibration for Long-Form Omni Hallucination

Long-video Omni assistants often fail not by inventing content, but by misbinding real evidence: they hear the right utterance and see the right event, yet attach it to the wrong speaker, moment, or modality. These \emph{almost-true} errors evade standard video QA because local evidence remains valid, so item-level scoring can reward both a supported claim and its near-counterfactual. We introduce a counterfactual event-binding protocol that constructs paired supported/counterfactual claims from the same audio-visual event evidence and evaluates them by strict-pair accuracy. We instantiate it as \bench, a benchmark for long-video Omni hallucination, with 3{,}600 single-claim QA items from 638 long-form videos averaging 24.16 minutes and covering 256.87 hours. Under this protocol, open-weight Omni models remain weak at pair-level binding: Qwen2.5-Omni-7B reaches 32.06\% and Qwen3-Omni-Instruct reaches 41.55\%, versus 76.54\% for a closed-source reference. To narrow this gap without updating the backbone, we propose \method, Modality-Perturbation Reliability Calibration, a frozen-backbone framework that selects audio-negative probes within video-level folds and fuses their response shifts with native audio-visual confidence into per-claim support estimates. \method lifts Qwen2.5-Omni-7B to 36.22\% and Qwen3 to 51.09\% on \bench, and improves target-adapted MCQ accuracy on OmniVideoBench ($+$2.20) and WorldSense ($+$1.51) with Qwen3.

cs.MM↗

arXiv2026

No Certificate, No Execution: Certified Traces as a Foundation for Trustworthy AI Agents

We argue that trustworthy AI agents, especially in high-stakes and policy-governed domains, should make execution conditional on certified traces rather than rely only on stronger generative models, output-level guardrails, or post-hoc audits. A generative agent may propose recommendations, tool calls, reports, or actions, but generation is not permission: an action may be computable yet impermissible, and individually permissible actions may compose into an impermissible trace. We formalize trustworthy agency through a \textbf{Proposal--Certification--Execution (PCE)} architecture: a probabilistic generating machine $M_G$ proposes candidate execution traces, a \textbf{Permissibility Machine} $M_Π$ certifies proposed traces under a policy system $Π$, and execution proceeds only for certified traces. The executable trace language is $L_{\mathrm{exec}} = L_G \cap L_{\mathrm{cert}}(M_Π)$. Before execution, a trace is a structured pre-execution record submitted for certification: it specifies intended steps, evidence, proposed tool calls, approvals, replayable computations, credentials, and execution conditions. This perspective complements chain-of-thought monitorability: visible reasoning may help detect misbehavior, but monitorability is not certifiability, and reasoning is only one component of a broader execution trace. The formal principle is simple: an agent-generated trace should execute only when it carries a checkable certificate witnessing permissibility under $Π$: \textbf{no certificate, no execution}. We develop certified traces and Permissibility Machines as foundations for trustworthy AI agents, connect trace certification to proof-carrying execution, proof memory, privacy, and zero-knowledge certificates, and propose evaluating agents by what generated traces can be safely certified for execution, not by output accuracy alone.

cs.CE↗

arXiv2026

Multi-SpatialMLLM: Multi-Frame Spatial Understanding with Multi-Modal Large Language Models

Multi-modal large language models (MLLMs) have rapidly advanced in visual tasks, yet their spatial understanding remains limited to single images, leaving them ill-suited for physical-world applications that require multi-frame reasoning. In this paper, we propose a framework to equip MLLMs with multi-frame spatial understanding by integrating fundamental spatial skills, including depth perception, visual correspondence, and dynamic perception. We design a novel data pipeline and collect the MultiSPA dataset of more than 27 million samples spanning diverse 3D and 4D scenes to enable training. Alongside MultiSPA, we introduce a comprehensive benchmark that tests a wide spectrum of spatial tasks under uniform metrics. Our resulting model, Multi-SpatialMLLM, achieves significant gains over baselines and proprietary systems, demonstrating scalable and generalizable multi-frame perception. We further observe multi-task benefits and emergent spatial capabilities in challenging scenarios, and showcase how our model can serve as a multi-frame reward annotator for robotics.

cs.CV↗

arXiv2026

Waveform Index Modulation for Backscatter Communications in RIS-Based MIMO Radars

This paper studies a symbiotic system in which a reconfigurable intelligent surface (RIS) assists a radar transmitter while conveying information to a reader via backscattering. The RIS is partitioned into subarrays that redirect the radar signal toward the angular sector under inspection and superimpose a slow-time modulation using orthogonal phase codes, thereby implementing MIMO radar functionalities. Communication is achieved by encoding information in the selection of an unordered subset of orthogonal codewords, without altering the RIS transmit beampattern. At the reader, the proposed index modulation scheme enables low-complexity detection without requiring channel state information. Numerical results demonstrate the effectiveness of the proposed backscatter communication approach.

eess.SP↗

arXiv2026

The Quantization Trap: Breaking Linear Scaling Laws in Multi-Hop Reasoning

Neural scaling laws provide a predictable recipe for AI advancement: reducing numerical precision should linearly improve computational efficiency and energy profile ($E \propto \mathrm{bits}$). In this paper, we demonstrate that this scaling law breaks in the context of multi-hop reasoning. We reveal a 'quantization trap' where reducing precision from 16-bit to 8/4-bit paradoxically increases net energy consumption while degrading reasoning accuracy. We provide a rigorous theoretical decomposition that attributes this failure to hardware casting overhead, the hidden latency cost of dequantization kernels, which becomes a dominant bottleneck in sequential reasoning chains, as well as to a sequential energy amortization failure. As a result, scaling law breaking is unavoidable in practice. We formalize a Critical Model Scale $N^*$ that predicts when the trap dissolves or deepens as a function of model size, batch size, and hardware configuration, validated across a 120$\times$ range (0.6B--72B) on six GPU architectures. Our findings suggest that the industry's "smaller-is-better" heuristic is mathematically counterproductive for complex reasoning tasks.

cs.AI↗

arXiv2026

Lyapunov spectrum of homoclinic classes

We study the Lyapunov spectrum of the ergodic measures of isolated homoclinic classes of $C^1$-generic diffeomorphisms. We show that this spectrum has nonempty interior and that any vector in its interior is the spectrum of some ergodic measure fully supported on the homoclinic class. We also discuss the averaged Lyapunov spectrum of homoclinic classes (an extension of the Lyapunov graph).

math.DS↗

arXiv2026

Direct and Ambient Backscatter Communications with a Dual-Function Radar Transmitter

This work considers a system where a dual-function radar transmitter (source) performs direct communication with a reader while simultaneously enabling ambient backscatter communication from a tag. The source embeds its message into a coded pulse repeatedly transmitted over a frame, whereas the tag exploits the resulting environmental reverberation (clutter) as an ambient carrier to convey its own message. By leveraging the structure induced by the radar waveforms, we develop two signaling schemes. In the pilot-free scheme, the source and tag messages are conveyed through nonlinear vector modulation; the induced subspace structure enables both joint decoding, where all unknown quantities are simultaneously estimated, and disjoint decoding, where the tag codeword is recovered first, followed by the estimation of the source codeword and the channel vectors. In the pilot-aided scheme, pilot symbols and linearly modulated data symbols are embedded within each frame, enabling both non-iterative decoding based on pilot-derived channel estimates and iterative decoding via alternating channel estimation and data detection. We establish sufficient conditions on the source and tag codebooks that guarantee noiseless identifiability of the involved messages and channels. Finally, performance is evaluated in terms of source/tag error probabilities and channel-estimation accuracy, and the resulting system-level tradeoffs are discussed.

eess.SP↗

arXiv2026

A Semi-amortized Lifted Learning-to-Optimize Masked (SALLO-M) Transformer Model for Scalable and Generalizable Beamforming

We develop an unsupervised deep learning framework for real-time scalable and generalizable downlink beamforming in multi-user multiple-input single-output (MU-MISO) systems. The proposed semi-amortized lifted learning-to-optimize (SALLO) framework employs a multi-layer Transformer to iteratively refine an auxiliary variable and the beamformer solution, with a few projected gradient ascent steps at each layer. A key feature of our SALLO Transformer model is that it can handle varying numbers of users and antennas, enabled by a user-antenna dual tokenization and a structured sample/attention masking scheme, leading to generalization across different configurations without retraining. To improve convergence and robustness, we introduce three training strategies: (a) sliding-window training to stabilize gradient propagation, (b) curriculum learning with random masking to enable user-antenna configuration generalization and prevent poor early-stage convergence, and (c) sample replay to mitigate catastrophic forgetting during multi-stage training. Ablation studies validate several key architecture designs and show that the enhanced training scheme improves both generalizability and solution quality. Simulation results over both Gaussian and sparse channels show that the proposed scheme consistently outperforms existing deep learning baselines across diverse system configurations and channel conditions. The performance gain becomes more pronounced in overloaded regimes, highlighting improved robustness under challenging scenarios. Furthermore, our scheme surpasses the WMMSE benchmark in underloaded systems and even in overloaded systems when the overloading factor is below certain threshold. These gains are achieved with fast inference and a substantially more lightweight model than wireless foundation models.

cs.LG↗

arXiv2026

Single-shot in situ pulse-duration measurement using plasma grating

Accurate measurement of the pulse duration of ultrashort, ultra-intense laser pulses at focus is essential for strong-field science. Most existing diagnostics, however, cannot allow direct in situ measurement in the focal region because of damage-threshold limits and unavoidable spatial averaging. We present a direct single-shot far-field diagnostic based on a plasma grating. In this method, the pulse duration is encoded in the axial length of an interference-written plasma grating and retrieved from the corresponding Bragg-diffraction signal. Comparison with near-field (pre-focus) autocorrelator measurements and far-field (at-focus) scanning measurements confirms single-shot pulse-duration retrieval in the focal region over 35-130 fs, and the method remains effective at a peak intensity of $\sim 10^{16}{\rm W/cm^2}$. In principle, the measurable range can be extended to 15-300 fs and to higher peak intensities. The method is insensitive to the laser central wavelength and offers a practical approach to far-field diagnostics in high-power laser systems.

physics.optics↗

arXiv2026

Resource Allocation for Positive-Rate Covert Communications Using Optimization and Deep Reinforcement Learning

We aim to achieve keyless covert communication with a positive-rate in Rayleigh block-fading channels. Specifically, the transmitter and the legitimate receiver are assumed to have either causal or non-causal knowledge of the \ac{CSI} for both the legitimate and the warden channels, while the warden only knows the statistical distribution of the \ac{CSI}. Two problem formulations are considered in this work: (a) Power allocation: maximizing the sum covert rate subject to a maximum power constraint, and (b) Rate allocation: minimizing the power consumption subject to a minimum covert rate constraint. Both problems are formulated based on recent information theoretical results on covert communication over state-dependent channels. When the \ac{CSI} of each fading block is known non-causally, we propose a novel three-step method to solve both the power and rate allocation problems. In the case where the \ac{CSI} is known causally, the power allocation problem can be formulated as \ac{MDP} and be solved using a \ac{DDQN} approach. Although the rate allocation problem under causal \ac{CSI} does not directly conform to an \ac{MDP} structure, it can be approximately solved using the \ac{DDQN} trained for power allocation. Simulation results demonstrate the effectiveness of the proposed power and rate allocation methods and provide comprehensive performance comparisons across different allocation schemes.

cs.IT↗

arXiv2026

MealRec: Multi-granularity Sequential Modeling via Hierarchical Diffusion Models for Micro-Video Recommendation

Micro-video recommendation aims to capture user preferences from the collaborative and context information of the interacted micro-videos, thereby predicting the appropriate videos. This target is often hindered by the inherent noise within multimodal content and unreliable implicit feedback, which weakens the correspondence between behaviors and underlying interests. While conventional works have predominantly approached such scenario through behavior-augmented modeling and content-centric multimodal analysis, these paradigms can inadvertently give rise to two non-trivial challenges: preference-irrelative video representation extraction and inherent modality conflicts. To address these issues, we propose a Multi-granularity sequential modeling method via hierarchical diffusion models for micro-video Recommendation (MealRec), which simultaneously considers temporal correlations during preference modeling from intra- and inter-video perspectives. Specifically, we first propose Temporal-guided Content Diffusion (TCD) to refine video representations under intra-video temporal guidance and personalized collaborative signals to emphasize salient content while suppressing redundancy. To achieve the semantically coherent preference modeling, we further design the Noise-unconditional Preference Denoising (NPD) to recovers informative user preferences from corrupted states under the blind denoising. Extensive experiments and analyses on four micro-video datasets from two platforms demonstrate the effectiveness, universality, and robustness of our MealRec, further uncovering the effective mechanism of our proposed TCD and NPD. The source code and corresponding dataset will be available upon acceptance.

cs.IR↗