arXiv subjects

Dong Yuan

Publications and source records attributed to Dong Yuan.

At least 19 recordsLinked to original sources

arXiv2026

WA-SpecDec: World-Aware Speculative Decoding for Vision-Language-Action Models

Vision-language-action (VLA) policies generate robot controls autoregressively, making closed-loop latency dominated by repeated target-model forward passes. Speculative decoding reduces this cost by verifying blocks of draft action tokens in parallel, and recent VLA methods further relax token-level acceptance because small differences in action-token space often map to similar continuous controls. However, this relaxation remains scene-agnostic. A fixed token-distance tolerance treats the same action-token deviation as equally safe across states, although deviations that are harmless in free space can cause collisions or grasp failures near contact. We propose WA-SpecDec, a world-aware speculative decoding framework that injects world-model-derived physical scene awareness during the VLA prefill stage, producing shared world-aware prefill states for draft proposal and target verification without changing the relaxed acceptance rule. Across three state-of-the-art relaxed acceptance schemes, WA-SpecDec preserves higher task success under looser relaxation and enables longer accepted prefixes. At comparable-success operating points, WA-SpecDec achieves a 1.5x matched-success speedup over VLA speculative decoding alone and reduces near-contact failure (NCF) by 18.6% on average relative to the corresponding speculative baselines.

cs.RO↗

arXiv2026

HorizonServe: Coordinating Request Scheduling with GPU Sharing for Omni-Model Serving

Omni models unify text, speech, image, and multimodal reasoning in a single serving backend, but this unified deployment exposes a new scheduling problem. Requests with different output modalities may share an initial multimodal backbone and then diverge into downstream generation stages, creating heterogeneous first-response metrics and service-level objective (SLO) targets on the same GPU. Existing large language model (LLM) and multimodal serving systems mainly optimize token progress or input-side processing, and they do not jointly control temporal sharing in the shared stage and spatial sharing among co-running stages. This paper presents HorizonServe, a single-GPU omni-model serving system that coordinates request admission and GPU allocation under heterogeneous SLOs. HorizonServe profiles per-class first-response latency, protects requests with limited slack, rotates shared-stage opportunities across execution paths, and throttles the shared-stage streaming multiprocessor (SM) allocation when downstream stages are active. Across three omni-model workloads and two GPU platforms, HorizonServe improves SLO attainment by up to 4.9$\times$ in arrival-rate sweeps and 7.0$\times$ under downstream-heavy traffic, and reduces per-class first-response latency by 38.4--63.7\%.

cs.DC↗

arXiv2026

mmRadarTwin: A Measurement-Calibrated Signal-Level Digital Twin Platform for Indoor mmWave Radar

Indoor mmWave radar perception is difficult to reproduce because measured range-angle responses depend on scene geometry, material response, multipath, hardware conventions, and signal processing. Existing ray-tracing and digital-twin tools often expose rendering, channel, or path-level quantities, while radar sensing requires complex signal products that can be processed and compared in the same domain as real FMCW measurements. We present mmRadarTwin, a signal-level and path-attributed digital-twin platform for indoor mmWave radar. mmRadarTwin links a real radar measurement branch with an Unreal Engine scene-simulation branch through a shared receive-channel and range-angle processing interface. The simulator writes complex multi-channel receive grids and exports per-path contribution records that identify the actor, material tag, propagation event, and output-bin support of each simulated return. We evaluate mmRadarTwin in an office deployment using a commodity monostatic mmWave radar and mobile scene-capture hardware. Across 154 measured poses spanning 22 radar locations, the current physics-only path-basis simulator recalls 70.8% of measurement-active geometry-supported response regions in the central usable field of view while exposing residuals caused by weak or missing path support, shifted responses, unsupported anchors, and missing physical mechanisms. Rather than claiming complete radar-map reconstruction or cross-room generalization, mmRadarTwin establishes a practical systems workflow for constructing, comparing, and diagnosing indoor radar digital twins.

cs.CV↗

arXiv2026

Understanding the Robustness of Distributed Self-Supervised Learning Frameworks Against Non-IID Data

Recent research has introduced distributed self-supervised learning (D-SSL) approaches to leverage vast amounts of unlabeled decentralized data. However, D-SSL faces the critical challenge of data heterogeneity, and there is limited theoretical understanding of how different D-SSL frameworks respond to this challenge. To fill this gap, we present a rigorous theoretical analysis of the robustness of D-SSL frameworks under non-IID (non-independent and identically distributed) settings. Our results show that pre-training with Masked Image Modeling (MIM) is inherently more robust to heterogeneous data than Contrastive Learning (CL), and that the robustness of decentralized SSL increases with average network connectivity, implying that federated learning (FL) is no less robust than decentralized learning (DecL). These findings provide a solid theoretical foundation for guiding the design of future D-SSL algorithms. To further illustrate the practical implications of our theory, we introduce MAR loss, a refinement of the MIM objective with local-to-global alignment regularization. Extensive experiments across model architectures and distributed settings validate our theoretical insights, and additionally confirm the effectiveness of MAR loss as an application of our analysis.

cs.LG↗

arXiv2026

PoseCompass: Intelligent Synthetic Pose Selection for Visual Localization

In visual localization, Absolute Pose Regression (APR) enables real-time 6-DoF camera pose inference from single images, yet critically depends on fine-tuning data quality and coverage. While recent methods leverage 3D Gaussian Splatting (3DGS) for novel view synthesis-based data augmentation, random sampling generates redundant views and noisy samples from poorly reconstructed regions. To mitigate this research gap, we propose PoseCompass, an intelligent pose selection pipeline for 3DGS-based APR. PoseCompass formulates synthetic pose selection and derives a value-based pose ranking mechanism to identify informative poses. The ranking integrates three dimensions: Localization Difficulty, favoring challenging regions; Coverage Novelty, exploring under-sampled areas; and Rendering Observability, filtering artifacts and noise. PoseCompass then generates trajectory-constrained candidates, selects the top-K ranked poses, and synthesizes views using 3DGS with lightweight diffusion-based alignment. Finally, the pose regressor is fine-tuned on mixed real and synthetic data. We evaluate PoseCompass on 7-Scenes, where it reduces adaptation time from 15.2 to 5.1 minutes, a 3x speedup, while cutting median pose errors by 53.8 percent and significantly outperforming random baselines.

cs.CV↗

arXiv2026

DuoServe-MoE: Dual-Phase Expert Prefetch and Caching for LLM Inference QoS Assurance

Large Language Models (LLMs) are increasingly deployed as Internet/Web services (LLM-as-a-Service) with strict latency Service-Level Objectives (SLOs) under tight GPU memory budgets. Mixture-of-Experts (MoE) models improve quality and throughput via sparse expert activation, but serving them efficiently is challenging because expert weights dominate memory footprint and incur costly host--device transfers when offloaded. Moreover, MoE serving exhibits a phase disparity: the prefill phase tends to activate experts densely across many tokens, while the decode phase activates only a few experts per step. A uniform expert loading/caching policy across phases leads to either peak-memory blowup (prefill) or tail-latency inflation (decode). We present DuoServe-MoE, a QoS-oriented MoE serving system that decouples prefill and decode and applies phase-specialized expert scheduling. For prefill, DuoServe-MoE uses a two-stream CUDA pipeline to overlap expert prefetching with non-MoE computation, reducing expert residency time and peak GPU memory. For decode, it employs a lightweight layer-level predictor trained offline from activation traces to prefetch only likely experts without model changes. Experiments on representative MoE LLMs show that DuoServe-MoE improves TTFT by up to $5.34\times$ and end-to-end latency by up to $7.55\times$ over representative baselines, while maintaining low runtime GPU memory usage under resource-constrained deployment.

cs.DC↗

arXiv2026

AgentServe: Algorithm-System Co-Design for Efficient Agentic AI Serving on a Consumer-Grade GPU

Large language models (LLMs) are increasingly deployed as AI agents that operate in short reasoning-action loops, interleaving model computation with external calls. Unlike traditional chat applications, these agentic workloads require inference serving systems to balance low latency, stable token emission, and throughput under multiple request arrivals from different AI agents. Recent deployments highlight a shift toward running small language models (SLMs) locally on consumer-grade GPUs, driven by privacy, compliance, and cost constraints. When heterogeneous requests overlap on a single GPU, long prefills and short decodes contend for resources, creating head-of-line blocking that destabilizes interactive performance. By analyzing agent workloads, we observe that their execution naturally separates into cold prefills, which process long system prompts, resume prefills, which append tool outputs to cached contexts, and short decodes, which are latency-critical. This mix intensifies contention compared to conventional chatbot serving. We present AgentServe, a single-GPU serving system that ensures stable multi-agent execution under such conditions by isolating prefills from decodes, applying dynamic budgeting to resume prefills, and allocating GPU resources through pre-established CUDA Green Context slots with adaptive control. Evaluation results show that AgentServe significantly improves latency stability while sustaining competitive throughput, achieving up to 2.8x TTFT improvement and 2.7x TPOT improvement over state-of-the-art baselines across different settings.

cs.DC↗

arXiv2026

Machine learning the arrow of time in solid-state spins

Understanding the emergence of the thermodynamic arrow of time in microscopic systems is of fundamental importance, particularly given that unitary evolution preserves time-reversal symmetry. While projective measurements introduce temporal irreversibility, identifying this asymmetry from single evolution trajectories in the presence of stochastic fluctuations presents a considerable challenge. Here, we harness machine learning to identify the arrow of time from individual trajectories generated by a programmable ten-qubit quantum processor based on a nitrogen-vacancy center in diamond. We implement quantum circuits that realize unitary evolutions where heat flows from hotter to colder subsystems and their time-reversed counterparts. Projective measurements inserted in these processes induce entropy production, and their outcomes constitute the evolution trajectory. We demonstrate that an unsupervised clustering algorithm autonomously divides the experimental trajectories into two distinct groups without prior knowledge, while a convolutional neural network identifies the temporal direction of these trajectories with approximately 92% accuracy. In addition, we show that a diffusion-based generative model reproduces essential signatures of directional energy flow and entropy production. Our results establish machine learning as a powerful tool for uncovering underlying physical processes from complex experimental data, advancing the interface between quantum thermodynamics and artificial intelligence.

quant-ph↗

arXiv2026

Optimal Look-back Horizon for Time Series Forecasting in Federated Learning

Selecting an appropriate look-back horizon remains a fundamental challenge in time series forecasting (TSF), particularly in the federated learning scenarios where data is decentralized, heterogeneous, and often non-independent. While recent work has explored horizon selection by preserving forecasting-relevant information in an intrinsic space, these approaches are primarily restricted to centralized and independently distributed settings. This paper presents a principled framework for adaptive horizon selection in federated time series forecasting through an intrinsic space formulation. We introduce a synthetic data generator (SDG) that captures essential temporal structures in client data, including autoregressive dependencies, seasonality, and trend, while incorporating client-specific heterogeneity. Building on this model, we define a transformation that maps time series windows into an intrinsic representation space with well-defined geometric and statistical properties. We then derive a decomposition of the forecasting loss into a Bayesian term, which reflects irreducible uncertainty, and an approximation term, which accounts for finite-sample effects and limited model capacity. Our analysis shows that while increasing the look-back horizon improves the identifiability of deterministic patterns, it also increases approximation error due to higher model complexity and reduced sample efficiency. We prove that the total forecasting loss is minimized at the smallest horizon where the irreducible loss starts to saturate, while the approximation loss continues to rise. This work provides a rigorous theoretical foundation for adaptive horizon selection for time series forecasting in federated learning.

cs.LG↗

arXiv2025

Scaling Law Analysis in Federated Learning: How to Select the Optimal Model Size?

The recent success of large language models (LLMs) has sparked a growing interest in training large-scale models. As the model size continues to scale, concerns are growing about the depletion of high-quality, well-curated training data. This has led practitioners to explore training approaches like Federated Learning (FL), which can leverage the abundant data on edge devices while maintaining privacy. However, the decentralization of training datasets in FL introduces challenges to scaling large models, a topic that remains under-explored. This paper fills this gap and provides qualitative insights on generalizing the previous model scaling experience to federated learning scenarios. Specifically, we derive a PAC-Bayes (Probably Approximately Correct Bayesian) upper bound for the generalization error of models trained with stochastic algorithms in federated settings and quantify the impact of distributed training data on the optimal model size by finding the analytic solution of model size that minimizes this bound. Our theoretical results demonstrate that the optimal model size has a negative power law relationship with the number of clients if the total training compute is unchanged. Besides, we also find that switching to FL with the same training compute will inevitably reduce the upper bound of generalization performance that the model can achieve through training, and that estimating the optimal model size in federated scenarios should depend on the average training compute across clients. Furthermore, we also empirically validate the correctness of our results with extensive training runs on different models, network settings, and datasets.

cs.LG↗

arXiv2025

GuardFed: A Trustworthy Federated Learning Framework Against Dual-Facet Attacks

Federated learning (FL) enables privacy-preserving collaborative model training but remains vulnerable to adversarial behaviors that compromise model utility or fairness across sensitive groups. While extensive studies have examined attacks targeting either objective, strategies that simultaneously degrade both utility and fairness remain largely unexplored. To bridge this gap, we introduce the Dual-Facet Attack (DFA), a novel threat model that concurrently undermines predictive accuracy and group fairness. Two variants, Synchronous DFA (S-DFA) and Split DFA (Sp-DFA), are further proposed to capture distinct real-world collusion scenarios. Experimental results show that existing robust FL defenses, including hybrid aggregation schemes, fail to resist DFAs effectively. To counter these threats, we propose GuardFed, a self-adaptive defense framework that maintains a fairness-aware reference model using a small amount of clean server data augmented with synthetic samples. In each training round, GuardFed computes a dual-perspective trust score for every client by jointly evaluating its utility deviation and fairness degradation, thereby enabling selective aggregation of trustworthy updates. Extensive experiments on real-world datasets demonstrate that GuardFed consistently preserves both accuracy and fairness under diverse non-IID and adversarial conditions, achieving state-of-the-art performance compared with existing robust FL methods.

cs.LG↗

arXiv2025

Time Independence Does Not Limit Information Flow. II. The Case with Ancillas

While the impact of locality restrictions on quantum dynamics and algorithmic complexity has been well studied in the general case of time-dependent Hamiltonians, the capabilities of time-independent protocols are less well understood. Using clock constructions, we show that the light cone for time-independent Hamiltonians, as captured by Lieb-Robinson bounds, is the same as that for time-dependent systems when local ancillas are allowed. More specifically, we develop time-independent protocols for approximate quantum state transfer with the same run-times as their corresponding time-dependent protocols. Given any piecewise-continuous Hamiltonian, our construction gives a time-independent Hamiltonian that implements its dynamics in the same time, up to error $\varepsilon$, at the cost of introducing a number of local ancilla qubits for each data qubit that is polylogarithmic in the number of qubits, the norm of the Hamiltonian and its derivative (if it exists), the run time, and $1/\varepsilon$. We apply this construction to state transfer for systems with power-law-decaying interactions and one-dimensional nearest-neighbor systems with disordered interaction strengths. In both cases, this gives time-independent protocols with the same optimal light-cone-saturating run-times as their time-dependent counterparts.

quant-ph↗

arXiv2025

Biology-Instructions: A Dataset and Benchmark for Multi-Omics Sequence Understanding Capability of Large Language Models

Large language models (LLMs) have shown remarkable capabilities in general domains, but their application to multi-omics biology remains underexplored. To address this gap, we introduce Biology-Instructions, the first large-scale instruction-tuning dataset for multi-omics biological sequences, including DNA, RNA, proteins, and multi-molecules. This dataset bridges LLMs and complex biological sequence-related tasks, enhancing their versatility and reasoning while maintaining conversational fluency. We also highlight significant limitations of current state-of-the-art LLMs on multi-omics tasks without specialized training. To overcome this, we propose ChatMultiOmics, a strong baseline with a novel three-stage training pipeline, demonstrating superior biological understanding through Biology-Instructions. Both resources are publicly available, paving the way for better integration of LLMs in multi-omics analysis. The Biology-Instructions is publicly available at: https://github.com/hhnqqq/Biology-Instructions.

q-bio.BM↗

arXiv2025

Real-time scattering and freeze-out dynamics in Rydberg-atom lattice gauge theory

Understanding the non-equilibrium dynamics of gauge theories remains a fundamental challenge in high-energy physics. Indeed, most large scale experiments on gauge theories intrinsically rely on very far-from equilibrium dynamics, from heavy-ion to lepton and hadron collisions, which is in general extremely challenging to treat ab initio. Quantum simulation holds intriguing potential in tackling this problem and pioneering experiments have observed different characteristic features of gauge theories, such as string breaking and false vacuum decay. Here, using a programmable Rydberg atom array, we observe real-time scattering and freeze-out dynamics in a (1+1)-dimensional U(1) lattice gauge theory. Through spatiotemporal Hamiltonian engineering, we demonstrate dynamical confinement-deconfinement transitions, revealing string fragmentation and symmetry restoration during quenches. We track scattering processes with single-site resolution across a range of parameter regimes. Utilizing a double quench protocol, we observe dynamical freeze-out: upon quenching the Hamiltonian after scattering, despite the injection of an extensive energy, the system evolution -- in terms of both low-order correlations and entanglement -- freezes, effectively stabilizing a highly correlated equilibrium state -- a situation that reminisces that of collisions between heavy ions. Our work establishes a high-resolution approach for probing non-perturbative gauge dynamics, opening alternative pathways toward studying far-from-equilibrium phenomena in high-energy physics.

cond-mat.quant-gas↗

arXiv2025

Dynamical freezing and enhanced magnetometry in an interacting spin ensemble

Understanding and controlling non-equilibrium dynamics in quantum many-body systems is a fundamental challenge in modern physics, with profound implications for advancing quantum technologies. Typically, periodically driven systems in the absence of conservation laws thermalize to a featureless "infinite-temperature" state, erasing all memory of their initial conditions. However, this paradigm can break down through mechanisms such as integrability, many-body localization, quantum many-body scars, and Hilbert space fragmentation. Here, we report the experimental observation of dynamical freezing, a distinct mechanism of thermalization breakdown in driven systems, and demonstrate its application in quantum sensing using an ensemble of approximately $10^4$ interacting nitrogen-vacancy spins in diamond. By precisely controlling the driving frequency and detuning, we observe emergent long-lived spin magnetization and coherent oscillatory micromotions, persisting over timescales exceeding the interaction-limited coherence time ($T_2$) by more than an order of magnitude. Leveraging these unconventional dynamics, we develop a dynamical-freezing-enhanced ac magnetometry that extends optimal sensing times far beyond $T_2$, outperforming conventional dynamical decoupling magnetometry with a 4.3 dB sensitivity enhancement. Our results not only provide clear experimental observation of dynamical freezing -- a peculiar mechanism defying thermalization through emergent conservation laws -- but also establish a robust control method generally applicable to diverse physical platforms, with broad implications in quantum metrology and beyond.

quant-ph↗

arXiv2025

PUMPS: Skeleton-Agnostic Point-based Universal Motion Pre-Training for Synthesis in Human Motion Tasks

Motion skeletons drive 3D character animation by transforming bone hierarchies, but differences in proportions or structure make motion data hard to transfer across skeletons, posing challenges for data-driven motion synthesis. Temporal Point Clouds (TPCs) offer an unstructured, cross-compatible motion representation. Though reversible with skeletons, TPCs mainly serve for compatibility, not for direct motion task learning. Doing so would require data synthesis capabilities for the TPC format, which presents unexplored challenges regarding its unique temporal consistency and point identifiability. Therefore, we propose PUMPS, the primordial autoencoder architecture for TPC data. PUMPS independently reduces frame-wise point clouds into sampleable feature vectors, from which a decoder extracts distinct temporal points using latent Gaussian noise vectors as sampling identifiers. We introduce linear assignment-based point pairing to optimise the TPC reconstruction process, and negate the use of expensive point-wise attention mechanisms in the architecture. Using these latent features, we pre-train a motion synthesis model capable of performing motion prediction, transition generation, and keyframe interpolation. For these pre-training tasks, PUMPS performs remarkably well even without native dataset supervision, matching state-of-the-art performance. When fine-tuned for motion denoising or estimation, PUMPS outperforms many respective methods without deviating from its generalist architecture.

cs.CV↗

arXiv2025

Motion Keyframe Interpolation for Any Human Skeleton via Temporally Consistent Point Cloud Sampling and Reconstruction

In the character animation field, modern supervised keyframe interpolation models have demonstrated exceptional performance in constructing natural human motions from sparse pose definitions. As supervised models, large motion datasets are necessary to facilitate the learning process; however, since motion is represented with fixed hierarchical skeletons, such datasets are incompatible for skeletons outside the datasets' native configurations. Consequently, the expected availability of a motion dataset for desired skeletons severely hinders the feasibility of learned interpolation in practice. To combat this limitation, we propose Point Cloud-based Motion Representation Learning (PC-MRL), an unsupervised approach to enabling cross-compatibility between skeletons for motion interpolation learning. PC-MRL consists of a skeleton obfuscation strategy using temporal point cloud sampling, and an unsupervised skeleton reconstruction method from point clouds. We devise a temporal point-wise K-nearest neighbors loss for unsupervised learning. Moreover, we propose First-frame Offset Quaternion (FOQ) and Rest Pose Augmentation (RPA) strategies to overcome necessary limitations of our unsupervised point cloud-to-skeletal motion process. Comprehensive experiments demonstrate the effectiveness of PC-MRL in motion interpolation for desired skeletons without supervision from native datasets.

cs.CV↗

arXiv2025

Time Independence Does Not Limit Information Flow. I. The Free-Particle Case

The speed of information propagation in long-range interacting quantum systems is limited by Lieb-Robinson-type bounds, whose tightness can be established by finding specific quantum state-transfer protocols. Previous works have given quantum state-transfer protocols that saturate the corresponding Lieb-Robinson bounds using time-dependent Hamiltonians. Are speed limits for quantum information propagation different for time-independent Hamiltonians? In a step towards addressing this question, we present and analyze two optimal time-independent state-transfer protocols for free-particle systems, which utilize continuous-time single-particle quantum walks with hopping strength decaying as a power law. We rigorously prove and numerically confirm that our protocols achieve quantum state transfer, with controllable error over an arbitrarily long distance in any spatial dimension, at the speed limits set by the free-particle Lieb-Robinson bounds. This shows that time independence does not limit information flow for long-range free-particle Hamiltonians.

quant-ph↗