arXiv subjects

Hong Zhang

Publications and source records attributed to Hong Zhang.

At least 73 records · Page 4Linked to original sources

arXiv2026

Improving Flow Matching by Aligning Flow Divergence

Conditional flow matching (CFM) stands out as an efficient, simulation-free approach for training flow-based generative models, achieving remarkable performance for data generation. However, CFM is insufficient to ensure accuracy in learning probability paths. In this paper, we introduce a new partial differential equation characterization for the error between the learned and exact probability paths, along with its solution. We show that the total variation gap between the two probability paths is bounded above by a combination of the CFM loss and an associated divergence loss. This theoretical insight leads to the design of a new objective function that simultaneously matches the flow and its divergence. Our new approach improves the performance of the flow-based generative model by a noticeable margin without sacrificing generation efficiency. We showcase the advantages of this enhanced training approach over CFM on several important benchmark tasks, including generative modeling for dynamical systems, DNA sequences, and videos. Code is available at \href{https://github.com/Utah-Math-Data-Science/Flow_Div_Matching}{Utah-Math-Data-Science}.

cs.LG↗

arXiv2026

Online Navigation Refinement: Achieving Lane-Level Guidance by Associating Standard-Definition and Online Perception Maps

Lane-level navigation is critical for geographic information systems and navigation-based tasks, offering finer-grained guidance than road-level navigation by standard definition (SD) maps. However, it currently relies on expansive global HD maps that cannot adapt to dynamic road conditions. Recently, online perception (OP) maps have become research hotspots, providing real-time geometry as an alternative, but lack the global topology needed for navigation. To address these issues, Online Navigation Refinement (ONR), a new mission is introduced that refines SD-map-based road-level routes into accurate lane-level navigation by associating SD maps with OP maps. The map-to-map association to handle many-to-one lane-to-road mappings under two key challenges: (1) no public dataset provides lane-to-road correspondences; (2) severe misalignment from spatial fluctuations, semantic disparities, and OP map noise invalidates traditional map matching. For these challenges, We contribute: (1) Online map association dataset (OMA), the first ONR benchmark with 30K scenarios and 2.6M annotated lane vectors; (2) MAT, a transformer with path-aware attention to aligns topology despite spatial fluctuations and semantic disparities and spatial attention for integrates noisy OP features via global context; and (3) NR P-R, a metric evaluating geometric and semantic alignment. Experiments show that MAT outperforms existing methods at 34 ms latency, enabling low-cost and up-to-date lane-level navigation.

cs.CV↗

arXiv2026

Quantitative FUP and spectral gap for quasi-Fuchsian group

We derive an explicit formula for the exponent $β$ in the higher-dimensional fractal uncertainty principle (FUP) established by Cohen 2023, quantifying its dependence on the porosity parameter $ν$ of the Fourier support. This quantitative version of FUP yields an explicit essential spectral gap for convex co-compact hyperbolic 3-manifolds arising from quasi-Fuchsian groups, thereby refining the result of Tao 2025. Our result extends the earlier work of Jin-Zhang 2020 to higher dimensions.

math.CA↗

arXiv2026

Local Group Velocity Distribution inside Superradiant Condensates

Superradiance enables scalar fields to extract energy and angular momentum from a rotating black hole (BH), leading to the formation of a BH-condensate system. Previous studies mainly focus on the phase velocity, which propagates in the azimuthal direction. In this work, we show that the superradiant scalar condensate presents a nontrivial group velocity distribution. In the region sufficiently far from the BH, the condensate exhibits a radial velocity magnitude that approaches $ (r_gμ/2) \sin (2ωt-2 φ)$, while the polar and azimuthal velocity magnitudes asymptotically decline as $\propto 1/r$.

gr-qc↗

arXiv2026

Effect of accretion on scalar superradiant instability

Superradiance can lead to the formation of a black hole (BH) condensate system. We thoroughly investigate the accretion effect on the evolution of this system, and the gravitational wave signals it emits in the presence of multiple superradiance modes. Assuming the multiplication of the BH mass and scalar mass as a small number, we obtain the analytical approximations of all important quantities, which can be directly applied to phenomenological studies. In addition, we confirm that accretion could significantly enhance the gravitational wave (GW) emission and reduce its duration, and show that the GW beat signature is similarly modified.

gr-qc↗

arXiv2025

Learnable Query Aggregation with KV Routing for Cross-view Geo-localisation

Cross-view geo-localisation (CVGL) aims to estimate the geographic location of a query image by matching it with images from a large-scale database. However, the significant view-point discrepancies present considerable challenges for effective feature aggregation and alignment. To address these challenges, we propose a novel CVGL system that incorporates three key improvements. Firstly, we leverage the DINOv2 backbone with a convolution adapter fine-tuning to enhance model adaptability to cross-view variations. Secondly, we propose a multi-scale channel reallocation module to strengthen the diversity and stability of spatial representations. Finally, we propose an improved aggregation module that integrates a Mixture-of-Experts (MoE) routing into the feature aggregation process. Specifically, the module dynamically selects expert subspaces for the keys and values in a cross-attention framework, enabling adaptive processing of heterogeneous input domains. Extensive experiments on the University-1652 and SUES-200 datasets demonstrate that our method achieves competitive performance with fewer trained parameters.

cs.CV↗

arXiv2025

Long-time stability and convergence analysis of an IMEX BDF3 scheme for 2-D incompressible Navier-Stokes equation

High-order time-stepping schemes are crucial for simulating incompressible fluid flows due to their ability to capture complex turbulent behavior and unsteady motion. In this work, we propose a third-order accurate numerical scheme for the two-dimensional incompressible Navier-Stokes equation. Spatial and temporal discretization is achieved using Fourier pseudo-spectral approximation and the BDF3 stencil, combined with the Adams-Bashforth extrapolation for the nonlinear convection term, resulting in a semi-implicit, fully discrete formulation. This approach requires solving only a single Poisson-like equation per time step while maintaining the desired temporal accuracy. Classical numerical experiments demonstrate the advantage of our scheme in terms of permissible time step sizes. Moreover, we establish uniform-in-time bounds for the vorticity in both $L^2$ and higher-order $H^m$ norms ($m \geq 1$), provided the time step is sufficiently small. These bounds, in turn, facilitate the derivation of optimal convergence rates.

math.NA↗

arXiv2025

Dexterous Manipulation through Imitation Learning: A Survey

Dexterous manipulation, which refers to the ability of a robotic hand or multi-fingered end-effector to skillfully control, reorient, and manipulate objects through precise, coordinated finger movements and adaptive force modulation, enables complex interactions similar to human hand dexterity. With recent advances in robotics and machine learning, there is a growing demand for these systems to operate in complex and unstructured environments. Traditional model-based approaches struggle to generalize across tasks and object variations due to the high dimensionality and complex contact dynamics of dexterous manipulation. Although model-free methods such as reinforcement learning (RL) show promise, they require extensive training, large-scale interaction data, and carefully designed rewards for stability and effectiveness. Imitation learning (IL) offers an alternative by allowing robots to acquire dexterous manipulation skills directly from expert demonstrations, capturing fine-grained coordination and contact dynamics while bypassing the need for explicit modeling and large-scale trial-and-error. This survey provides an overview of dexterous manipulation methods based on imitation learning, details recent advances, and addresses key challenges in the field. Additionally, it explores potential research directions to enhance IL-driven dexterous manipulation. Our goal is to offer researchers and practitioners a comprehensive introduction to this rapidly evolving domain.

cs.RO↗

arXiv2025

Monocular Person Localization under Camera Ego-motion

Localizing a person from a moving monocular camera is critical for Human-Robot Interaction (HRI). To estimate the 3D human position from a 2D image, existing methods either depend on the geometric assumption of a fixed camera or use a position regression model trained on datasets containing little camera ego-motion. These methods are vulnerable to severe camera ego-motion, resulting in inaccurate person localization. We consider person localization as a part of a pose estimation problem. By representing a human with a four-point model, our method jointly estimates the 2D camera attitude and the person's 3D location through optimization. Evaluations on both public datasets and real robot experiments demonstrate our method outperforms baselines in person localization accuracy. Our method is further implemented into a person-following system and deployed on an agile quadruped robot.

cs.CV↗

arXiv2025

MfNeuPAN: Proactive End-to-End Navigation in Dynamic Environments via Direct Multi-Frame Point Constraints

Obstacle avoidance in complex and dynamic environments is a critical challenge for real-time robot navigation. Model-based and learning-based methods often fail in highly dynamic scenarios because traditional methods assume a static environment and cannot adapt to real-time changes, while learning-based methods rely on single-frame observations for motion constraint estimation, limiting their adaptability. To overcome these limitations, this paper proposes a novel framework that leverages multi-frame point constraints, including current and future frames predicted by a dedicated module, to enable proactive end-to-end navigation. By incorporating a prediction module that forecasts the future path of moving obstacles based on multi-frame observations, our method allows the robot to proactively anticipate and avoid potential dangers. This proactive planning capability significantly enhances navigation robustness and efficiency in unknown dynamic environments. Simulations and real-world experiments validate the effectiveness of our approach.

cs.RO↗

arXiv2025

Multispectral-NeRF:a multispectral modeling approach based on neural radiance fields

3D reconstruction technology generates three-dimensional representations of real-world objects, scenes, or environments using sensor data such as 2D images, with extensive applications in robotics, autonomous vehicles, and virtual reality systems. Traditional 3D reconstruction techniques based on 2D images typically relies on RGB spectral information. With advances in sensor technology, additional spectral bands beyond RGB have been increasingly incorporated into 3D reconstruction workflows. Existing methods that integrate these expanded spectral data often suffer from expensive scheme prices, low accuracy and poor geometric features. Three - dimensional reconstruction based on NeRF can effectively address the various issues in current multispectral 3D reconstruction methods, producing high - precision and high - quality reconstruction results. However, currently, NeRF and some improved models such as NeRFacto are trained on three - band data and cannot take into account the multi - band information. To address this problem, we propose Multispectral-NeRF, an enhanced neural architecture derived from NeRF that can effectively integrates multispectral information. Our technical contributions comprise threefold modifications: Expanding hidden layer dimensionality to accommodate 6-band spectral inputs; Redesigning residual functions to optimize spectral discrepancy calculations between reconstructed and reference images; Adapting data compression modules to address the increased bit-depth requirements of multispectral imagery. Experimental results confirm that Multispectral-NeRF successfully processes multi-band spectral features while accurately preserving the original scenes' spectral characteristics.

cs.CV↗

arXiv2025

FullPart: Generating each 3D Part at Full Resolution

Part-based 3D generation holds great potential for various applications. Previous part generators that represent parts using implicit vector-set tokens often suffer from insufficient geometric details. Another line of work adopts an explicit voxel representation but shares a global voxel grid among all parts; this often causes small parts to occupy too few voxels, leading to degraded quality. In this paper, we propose FullPart, a novel framework that combines both implicit and explicit paradigms. It first derives the bounding box layout through an implicit box vector-set diffusion process, a task that implicit diffusion handles effectively since box tokens contain little geometric detail. Then, it generates detailed parts, each within its own fixed full-resolution voxel grid. Instead of sharing a global low-resolution space, each part in our method - even small ones - is generated at full resolution, enabling the synthesis of intricate details. We further introduce a center-point encoding strategy to address the misalignment issue when exchanging information between parts of different actual sizes, thereby maintaining global coherence. Moreover, to tackle the scarcity of reliable part data, we present PartVerse-XL, the largest human-annotated 3D part dataset to date with 40K objects and 320K parts. Extensive experiments demonstrate that FullPart achieves state-of-the-art results in 3D part generation. We will release all code, data, and model to benefit future research in 3D part generation.

cs.CV↗

arXiv2025

CogStereo: Neural Stereo Matching with Implicit Spatial Cognition Embedding

Deep stereo matching has advanced significantly on benchmark datasets through fine-tuning but falls short of the zero-shot generalization seen in foundation models in other vision tasks. We introduce CogStereo, a novel framework that addresses challenging regions, such as occlusions or weak textures, without relying on dataset-specific priors. CogStereo embeds implicit spatial cognition into the refinement process by using monocular depth features as priors, capturing holistic scene understanding beyond local correspondences. This approach ensures structurally coherent disparity estimation, even in areas where geometry alone is inadequate. CogStereo employs a dual-conditional refinement mechanism that combines pixel-wise uncertainty with cognition-guided features for consistent global correction of mismatches. Extensive experiments on Scene Flow, KITTI, Middlebury, ETH3D, EuRoc, and real-world demonstrate that CogStereo not only achieves state-of-the-art results but also excels in cross-domain generalization, shifting stereo vision towards a cognition-driven approach.

cs.CV↗

arXiv2025

Light Axion-Like Particles at Future Lepton Colliders

Axion-like particles (ALPs) are well-motivated extensions of the Standard Model (SM) that appear in many new physics scenarios, with masses spanning a broad range. In this work, we systematically study the production and detection prospects of light ALPs at future lepton colliders, including electron-positron and multi-TeV muon colliders. At lepton colliders, light ALPs can be produced in association with a photon or a $Z$ boson. For very light ALPs ($m_a < 1$ MeV), the ALPs are typically long-lived and escape detection, leading to a mono-$V$ ($V = γ, Z$) signature. In the long-lived limit, we find that the mono-photon channel at the Tera-$Z$ stage of future electron-positron colliders provides the strongest constraints on ALP couplings to SM gauge bosons, $g_{aVV}$, thanks to the high luminosity, low background, and resonant enhancement from on-shell $Z$ bosons. At higher energies, the mono-photon cross section becomes nearly energy-independent, and the sensitivity is governed by luminosity and background. At multi-TeV muon colliders, the mono-$Z$ channel can yield complementary constraints. For heavier ALPs ($m_a > 100$ MeV) that decay promptly, mono-$V$ signatures are no longer valid. In this case, ALPs can be probed via non-resonant vector boson scattering (VBS) processes, where the ALP is exchanged off-shell, leading to kinematic deviations from SM expectations. We analyze constraints from both light-by-light scattering and electroweak VBS, the latter only accessible at TeV-scale colliders. While generally weaker, these constraints are robust and model-independent. Our combined analysis shows that mono-$V$ and non-resonant VBS channels provide powerful and complementary probes of ALP-gauge boson interactions.

hep-ph↗

arXiv2025

Adap-RPF: Adaptive Trajectory Sampling for Robot Person Following in Dynamic Crowded Environments

Robot person following (RPF) is a core capability in human-robot interaction, enabling robots to assist users in daily activities, collaborative work, and other service scenarios. However, achieving practical RPF remains challenging due to frequent occlusions, particularly in dynamic and crowded environments. Existing approaches often rely on fixed-point following or sparse candidate-point selection with oversimplified heuristics, which cannot adequately handle complex occlusions caused by moving obstacles such as pedestrians. To address these limitations, we propose an adaptive trajectory sampling method that generates dense candidate points within socially aware zones and evaluates them using a multi-objective cost function. Based on the optimal point, a person-following trajectory is estimated relative to the predicted motion of the target. We further design a prediction-aware model predictive path integral (MPPI) controller that simultaneously tracks this trajectory and proactively avoids collisions using predicted pedestrian motions. Extensive experiments show that our method outperforms state-of-the-art baselines in smoothness, safety, robustness, and human comfort, with its effectiveness further demonstrated on a mobile robot in real-world scenarios.

cs.RO↗

arXiv2025

TrackVLA++: Unleashing Reasoning and Memory Capabilities in VLA Models for Embodied Visual Tracking

Embodied Visual Tracking (EVT) is a fundamental ability that underpins practical applications, such as companion robots, guidance robots and service assistants, where continuously following moving targets is essential. Recent advances have enabled language-guided tracking in complex and unstructured scenes. However, existing approaches lack explicit spatial reasoning and effective temporal memory, causing failures under severe occlusions or in the presence of similar-looking distractors. To address these challenges, we present TrackVLA++, a novel Vision-Language-Action (VLA) model that enhances embodied visual tracking with two key modules, a spatial reasoning mechanism and a Target Identification Memory (TIM). The reasoning module introduces a Chain-of-Thought paradigm, termed Polar-CoT, which infers the target's relative position and encodes it as a compact polar-coordinate token for action prediction. Guided by these spatial priors, the TIM employs a gated update strategy to preserve long-horizon target memory, ensuring spatiotemporal consistency and mitigating target loss during extended occlusions. Extensive experiments show that TrackVLA++ achieves state-of-the-art performance on public benchmarks across both egocentric and multi-camera settings. On the challenging EVT-Bench DT split, TrackVLA++ surpasses the previous leading approach by 5.1 and 12, respectively. Furthermore, TrackVLA++ exhibits strong zero-shot generalization, enabling robust real-world tracking in dynamic and occluded scenarios.

cs.RO↗

arXiv2025

Taming Latency-Memory Trade-Off in MoE-Based LLM Serving via Fine-Grained Expert Offloading

Large Language Models (LLMs) have gained immense success in revolutionizing various applications, including content generation, search and recommendation, and AI-assisted operation. To reduce high training costs, Mixture-of-Experts (MoE) architecture has become a popular backbone for modern LLMs. However, despite the benefits, serving MoE-based LLMs experience severe memory inefficiency due to sparsely activated experts. Recent studies propose to offload inactive experts from GPU memory to CPU memory to improve the serving efficiency of MoE models. However, they either incur high inference latency or high model memory footprints due to coarse-grained designs. To tame the latency-memory trade-off in MoE serving, we present FineMoE, a fine-grained expert offloading system for MoE serving that achieves low inference latency with memory efficiency. We design FineMoE to extract fine-grained expert selection patterns from MoE models and semantic hints from input prompts to efficiently guide expert prefetching, caching, and offloading decisions. FineMoE is prototyped on top of HuggingFace Transformers and deployed on a six-GPU testbed. Experiments with open-source MoE models and real-world workloads show that FineMoE reduces inference latency by 47% and improves expert hit rate by 39% over state-of-the-art solutions.

cs.LG↗

arXiv2025

Superradiant dark matter production from primordial black holes: Impact of multiple modes and gravitational wave emission

Rotating primordial black holes (PBHs) in the early universe can emit particles through superradiance, a process particularly efficient when the particle's Compton wavelength is comparable to the PBH's gravitational radius. Superradiance leads to an exponential growth of particle occupation numbers in gravitationally bound states. We present an analysis of heavy bosonic dark matter (DM) production through three gravitational mechanisms: Hawking radiation, superradiant instabilities, and ultraviolet (UV) freeze-in. We consider PBHs that evaporate before Big Bang Nucleosynthesis (BBN). For both scalar and vector DM, our analysis incorporates the evolution of a second superradiant mode. We demonstrate that the growth of a second superradiant mode causes the decay of the first mode, and thus the second mode cannot further enhance the DM abundance beyond that already achieved by the first mode. Our study also reveals that while superradiance generally enhances DM production, gravitational wave (GW) emission from the superradiant cloud may significantly modify this picture. For scalar DM, GW emission reduces the parameter space where superradiance effectively augments relic abundance. For vector DM, rapid GW emission from the superradiant cloud may yield relic abundances below those achieved through Hawking radiation alone. These findings demonstrate that multiple-mode effect and GW emission play critical roles in modeling DM production from PBHs in the early universe.

astro-ph.CO↗