arXiv subjects

Hao Xu

Publications and source records attributed to Hao Xu.

At least 199 records · Page 11Linked to original sources

arXiv2025

Long-range Meta-path Search on Large-scale Heterogeneous Graphs

Utilizing long-range dependency, a concept extensively studied in homogeneous graphs, remains underexplored in heterogeneous graphs, especially on large ones, posing two significant challenges: Reducing computational costs while maximizing effective information utilization in the presence of heterogeneity, and overcoming the over-smoothing issue in graph neural networks. To address this gap, we investigate the importance of different meta-paths and introduce an automatic framework for utilizing long-range dependency on heterogeneous graphs, denoted as Long-range Meta-path Search through Progressive Sampling (LMSPS). Specifically, we develop a search space with all meta-paths related to the target node type. By employing a progressive sampling algorithm, LMSPS dynamically shrinks the search space with hop-independent time complexity. Through a sampling evaluation strategy, LMSPS conducts a specialized and effective meta-path selection, leading to retraining with only effective meta-paths, thus mitigating costs and over-smoothing. Extensive experiments across diverse heterogeneous datasets validate LMSPS's capability in discovering effective long-range meta-paths, surpassing state-of-the-art methods. Our code is available at https://github.com/JHL-HUST/LMSPS.

cs.AI↗

arXiv2025

UniHOPE: A Unified Approach for Hand-Only and Hand-Object Pose Estimation

Estimating the 3D pose of hand and potential hand-held object from monocular images is a longstanding challenge. Yet, existing methods are specialized, focusing on either bare-hand or hand interacting with object. No method can flexibly handle both scenarios and their performance degrades when applied to the other scenario. In this paper, we propose UniHOPE, a unified approach for general 3D hand-object pose estimation, flexibly adapting both scenarios. Technically, we design a grasp-aware feature fusion module to integrate hand-object features with an object switcher to dynamically control the hand-object pose estimation according to grasping status. Further, to uplift the robustness of hand pose estimation regardless of object presence, we generate realistic de-occluded image pairs to train the model to learn object-induced hand occlusions, and formulate multi-level feature enhancement techniques for learning occlusion-invariant features. Extensive experiments on three commonly-used benchmarks demonstrate UniHOPE's SOTA performance in addressing hand-only and hand-object scenarios. Code will be released on https://github.com/JoyboyWang/UniHOPE_Pytorch.

cs.CV↗

arXiv2025

Higher Witt Groups for 2-Categories I: Centralizers

In this article, we investigate monoidal, braided, sylleptic centralizers of monoidal, braided, sylleptic 2-functors. We specifically focus on multifusion 2-categories and show that monoidal, braided, sylleptic centralizers are multifusion again, via studying the corresponding enveloping algebras. We provide a characterization of the non-degeneracy condition for monoidal, braided, and sylleptic fusion 2-categories, via vanishing of their centers. Applying Double Centralizer Theorems, we establish the relationship between monoidal, braided, symmetric local modules and free modules. In particular, we obtain factorization properties of non-degenerate monoidal, braided, and sylleptic fusion 2-categories. Main results in this article will be used to study higher Witt equivalences of non-degenerate monoidal, braided, sylleptic 2-categories in the sequential articles.

math.CT↗

arXiv2025

Dynamic Manipulation of Multiphase Fluid in Microgravity Using Photoresponsive Surfactant

Control of bubble motion is essential for improving efficiency and creating new functionalities in electrochemistry, heat transfer, and biomedical systems. Photoresponsive surfactants enable bubble manipulation by creating surface tension gradients, inducing a photo-Marangoni flow under illumination, without needing any engineered substrates, by leveraging a reversible switch in molecular conformation. Although previous studies have demonstrated bubble manipulation using photo-responsive surfactants, a comprehensive understanding of how fluid behavior is affected by critical parameters, such as bubble size, illumination, photo-switching kinetics, concentration, and adsorption desorption kinetics, remains elusive. Advances have been limited by the complex multiphysics processed involved, and by the fact that earth-bound experiments cannot study bubble photo-Marangoni dynamics without interference from bubble buoyancy and photo-thermal convection. We elucidate the factors enabling fast photo-Marangoni-driven bubble motion, by performing microgravity experiments, enabled by a bespoke photo-surfactant, complemented by a detailed modeling framework. We identify an optimal bubble size for migration, since smaller and larger bubbles incur weaker photo-Marangoni stresses and larger drag, respectively. Surfactants that switch rapidly under illumination drive fast migration, provided their reverse switch (in darkness) is much slower, yet not negligible. These foundational results enable the synthesis of next-generation photo-surfactants and photo-Marangoni manipulation across multiphase fluid systems.

physics.flu-dyn↗

arXiv2025

HybridReg: Robust 3D Point Cloud Registration with Hybrid Motions

Scene-level point cloud registration is very challenging when considering dynamic foregrounds. Existing indoor datasets mostly assume rigid motions, so the trained models cannot robustly handle scenes with non-rigid motions. On the other hand, non-rigid datasets are mainly object-level, so the trained models cannot generalize well to complex scenes. This paper presents HybridReg, a new approach to 3D point cloud registration, learning uncertainty mask to account for hybrid motions: rigid for backgrounds and non-rigid/rigid for instance-level foregrounds. First, we build a scene-level 3D registration dataset, namely HybridMatch, designed specifically with strategies to arrange diverse deforming foregrounds in a controllable manner. Second, we account for different motion types and formulate a mask-learning module to alleviate the interference of deforming outliers. Third, we exploit a simple yet effective negative log-likelihood loss to adopt uncertainty to guide the feature extraction and correlation computation. To our best knowledge, HybridReg is the first work that exploits hybrid motions for robust point cloud registration. Extensive experiments show HybridReg's strengths, leading it to achieve state-of-the-art performance on both widely-used indoor and outdoor datasets.

cs.CV↗

arXiv2025

MultiCo3D: Multi-Label Voxel Contrast for One-Shot Incremental Segmentation of 3D Neuroimages

3D neuroimages provide a comprehensive view of brain structure and function, aiding in precise localization and functional connectivity analysis. Segmentation of white matter (WM) tracts using 3D neuroimages is vital for understanding the brain's structural connectivity in both healthy and diseased states. One-shot Class Incremental Semantic Segmentation (OCIS) refers to effectively segmenting new (novel) classes using only a single sample while retaining knowledge of old (base) classes without forgetting. Voxel-contrastive OCIS methods adjust the feature space to alleviate the feature overlap problem between the base and novel classes. However, since WM tract segmentation is a multi-label segmentation task, existing single-label voxel contrastive-based methods may cause inherent contradictions. To address this, we propose a new multi-label voxel contrast framework called MultiCo3D for one-shot class incremental tract segmentation. Our method utilizes uncertainty distillation to preserve base tract segmentation knowledge while adjusting the feature space with multi-label voxel contrast to alleviate feature overlap when learning novel tracts and dynamically weighting multi losses to balance overall loss. We compare our method against several state-of-the-art (SOTA) approaches. The experimental results show that our method significantly enhances one-shot class incremental tract segmentation accuracy across five different experimental setups on HCP and Preto datasets.

cs.CV↗

arXiv2025

Fluid Antenna System Empowering 5G NR

Fluid antenna system (FAS) is an emerging technology that uses the new form of shape- and position-reconfigurable antennas to empower the physical layer for wireless communications. Prior studies on FAS were however limited to narrowband channels. Motivated by this, this paper addresses the integration of FAS in the fifth generation (5G) orthogonal frequency division multiplexing (OFDM) framework to address the challenges posed by wideband communications. We propose the framework of the wideband FAS OFDM system that includes a novel port selection matrix. Then we derive the achievable rate expression and design the adaptive modulation and coding (AMC) scheme based on the rate. Extensive link-level simulation results demonstrate striking improvements of FAS in the wideband channels, underscoring the potential of FAS in future wireless communications.

eess.SP↗

arXiv2025

Applying Deep Learning to Ads Conversion Prediction in Last Mile Delivery Marketplace

Deep neural networks (DNNs) have revolutionized web-scale ranking systems, enabling breakthroughs in capturing complex user behaviors and driving performance gains. At DoorDash, we first harnessed this transformative power by transitioning our homepage Ads ranking system from traditional tree based models to cutting edge multi task DNNs. This evolution sparked advancements in data foundations, model design, training efficiency, evaluation rigor, and online serving, delivering substantial business impact and reshaping our approach to machine learning. In this paper, we talk about our problem driven journey, from identifying the right problems and crafting targeted solutions to overcoming the complexity of developing and scaling a deep learning recommendation system. Through our successes and learned lessons, we aim to share insights and practical guidance to teams pursuing similar advancements in machine learning systems.

cs.LG↗

arXiv2025

How to Alleviate Catastrophic Forgetting in LLMs Finetuning? Hierarchical Layer-Wise and Element-Wise Regularization

Large Language Models (LLMs) exhibit strong general language capabilities. However, fine-tuning these models on domain-specific tasks often leads to catastrophic forgetting, where the model overwrites or loses essential knowledge acquired during pretraining. This phenomenon significantly limits the broader applicability of LLMs. To address this challenge, we propose a novel approach to compute the element-wise importance of model parameters crucial for preserving general knowledge during fine-tuning. Our method utilizes a dual-objective optimization strategy: (1) regularization loss based on element-wise parameter importance, which constrains the updates to parameters crucial for general knowledge; (2) cross-entropy loss to adapt to domain-specific tasks. Additionally, we introduce layer-wise coefficients to account for the varying contributions of different layers, dynamically balancing the dual-objective optimization. Extensive experiments on scientific, medical, and physical tasks using GPT-J and LLaMA-3 demonstrate that our approach mitigates catastrophic forgetting while enhancing model adaptability. Compared to previous methods, our solution is approximately 20 times faster and requires only 10-15% of the storage, highlighting the practical efficiency. The code will be released.

cs.CL↗

arXiv2025

Medical Image Registration Meets Vision Foundation Model: Prototype Learning and Contour Awareness

Medical image registration is a fundamental task in medical image analysis, aiming to establish spatial correspondences between paired images. However, existing unsupervised deformable registration methods rely solely on intensity-based similarity metrics, lacking explicit anatomical knowledge, which limits their accuracy and robustness. Vision foundation models, such as the Segment Anything Model (SAM), can generate high-quality segmentation masks that provide explicit anatomical structure knowledge, addressing the limitations of traditional methods that depend only on intensity similarity. Based on this, we propose a novel SAM-assisted registration framework incorporating prototype learning and contour awareness. The framework includes: (1) Explicit anatomical information injection, where SAM-generated segmentation masks are used as auxiliary inputs throughout training and testing to ensure the consistency of anatomical information; (2) Prototype learning, which leverages segmentation masks to extract prototype features and aligns prototypes to optimize semantic correspondences between images; and (3) Contour-aware loss, a contour-aware loss is designed that leverages the edges of segmentation masks to improve the model's performance in fine-grained deformation fields. Extensive experiments demonstrate that the proposed framework significantly outperforms existing methods across multiple datasets, particularly in challenging scenarios with complex anatomical structures and ambiguous boundaries. Our code is available at https://github.com/HaoXu0507/IPMI25-SAM-Assisted-Registration.

cs.CV↗

arXiv2025

Significant Chiral Magnetotransport Magnified by Multiple Weyl Nodes

The intertwining of magnetism with topology is known to give rise to exotic quantum phenomena. Here, we explore the magnetotransport properties of NdAlSi, a magnetic Weyl semimetal that spontaneously breaks inversion and time-reversal symmetries and hosts a large number of Weyl nodes. We observe a significant negative magnetoresistance, which we attribute to the chiral anomaly associated with multiple Weyl nodes. Remarkably, the extracted chiral coefficient reaches approximately $52~\mathrm{mΩ}^{-1}~\mathrm{m}^{-1}~\mathrm{T}^{-2}$, larger than many other topological materials. Additionally, we observe an exotic anomalous Hall effect with an out-of-sync behavior, where the anomalous Hall resistance does not exactly follow the field dependence of the magnetization, in contrast to that in conventional ferromagnets. These rich quantum transport phenomena, driven by the interplay between magnetism and Weyl nodes, establish NdAlSi as a prime platform for exploring the intricate topological behaviors of magnetic Weyl semimetals.

cond-mat.mes-hall↗

arXiv2025

DiversityOne: A Multi-Country Smartphone Sensor Dataset for Everyday Life Behavior Modeling

Understanding everyday life behavior of young adults through personal devices, e.g., smartphones and smartwatches, is key for various applications, from enhancing the user experience in mobile apps to enabling appropriate interventions in digital health apps. Towards this goal, previous studies have relied on datasets combining passive sensor data with human-provided annotations or self-reports. However, many existing datasets are limited in scope, often focusing on specific countries primarily in the Global North, involving a small number of participants, or using a limited range of pre-processed sensors. These limitations restrict the ability to capture cross-country variations of human behavior, including the possibility of studying model generalization, and robustness. To address this gap, we introduce DiversityOne, a dataset which spans eight countries (China, Denmark, India, Italy, Mexico, Mongolia, Paraguay, and the United Kingdom) and includes data from 782 college students over four weeks. DiversityOne contains data from 26 smartphone sensor modalities and 350K+ self-reports. As of today, it is one of the largest and most diverse publicly available datasets, while featuring extensive demographic and psychosocial survey data. DiversityOne opens the possibility of studying important research problems in ubiquitous computing, particularly in domain adaptation and generalization across countries, all research areas so far largely underexplored because of the lack of adequate datasets.

cs.CY↗

arXiv2025

Fast Inventory for 3GPP Ambient IoT Considering Device Unavailability due to Energy Harvesting

With the growing demand for massive internet of things (IoT), new IoT technology, namely ambient IoT (A-IoT), has been studied in the 3rd Generation Partnership Project (3GPP). A-IoT devices are batteryless and consume ultra-low power, relying on energy harvesting and energy storage to capture a small amount of energy for communication. A promising usecase of A-IoT is inventory, where a reader communicates with hundreds of A-IoT devices to identify them. However, energy harvesting required before communication can significantly delay or even fail inventory completion. In this work, solutions including duty cycled monitoring (DCM), device grouping and low-power receiving chain are proposed. Evaluation results show that the time required for a reader to complete an inventory procedure for hundreds of A-IoT devices can be reduced by 50% to 83% with the proposed methods.

eess.SP↗

arXiv2025

Low-Complex Waveform, Modulation and Coding Designs for 3GPP Ambient IoT

This paper presents a comprehensive study on low-complexity waveform, modulation and coding (WMC) designs for the 3rd Generation Partnership Project (3GPP) Ambient Internet of Things (A-IoT). A-IoT is a low-cost, low-power IoT system inspired by Ultra High Frequency (UHF) Radio Frequency Identification (RFID) and aims to leverage existing cellular network infrastructure for efficient RF tag management. The paper compares the physical layer (PHY) design challenges and requirements of RFID and A-IoT, particularly focusing on backscatter communications. An overview of the standardization for PHY designs in Release 19 A-IoT is provided, along with detailed schemes of the proposed low-complex WMC designs. The performance of device-to-reader link designs is validated through simulations, demonstrating 6 dB improvements of the proposed baseband waveform with coherent receivers compared to RFID line coding-based solutions with non-coherent receivers when channel coding is adopted.

eess.SP↗

arXiv2025

Magnetic Interactions in the Polar Ferrimagnet with a Bipartite Structure

The polar magnets A$_2$Mo$_3$O$_8$ (A=Fe, Mn, Co, and Ni) feature a bipartite structure, where the magnetic A$^{2+}$ ions occupy two different sites with octahedral and tetrahedral oxygen coordinations. This bipartite structure provides a platform for the emergence of nontrivial magnetoelectric (ME) effects and intriguing excitation behaviors, and thus creates significant research interest. In this study, we conduct inelastic neutron scattering measurements on single crystals of Mn$_2$Mo$_3$O$_8$, an L-type ferrimagnet in the A$_2$Mo$_3$O$_8$ family, to investigate its spin dynamics. The obtained magnetic excitation spectra reveal two distinct magnon dispersions corresponding to the octahedral and tetrahedral spins in Mn$_2$Mo$_3$O$_8$. These magnon bands can be well described by a spin Hamiltonian including Heisenberg and single-ion anisotropy terms. Employing our effective spin model, we successfully reproduce the unusual temperature dependence of the L-type ferrimagnetic susceptibility through self-consistent mean-field theory. This research reveals the significance of the bipartite structure in determining the excitation properties of the polar magnets $\rm{A_{2}Mo_{3}O_{8}}$ and provides valuable insights into the spin dynamics of L-type ferrimagnets.

cond-mat.str-el↗

arXiv2025

Downlink OFDM-FAMA in 5G-NR Systems

Fluid antenna multiple access (FAMA), enabled by the fluid antenna system (FAS), offers a new and straightforward solution to massive connectivity. Previous results on FAMA were primarily based on narrowband channels. This paper studies the adoption of FAMA within the fifth-generation (5G) orthogonal frequency division multiplexing (OFDM) framework, referred to as OFDM-FAMA, and evaluate its performance in broadband multipath channels. We first design the OFDM-FAMA system, taking into account 5G channel coding and OFDM modulation. Then the system's achievable rate is analyzed, and an algorithm to approximate the FAS configuration at each user is proposed based on the rate. Extensive link-level simulation results reveal that OFDM-FAMA can significantly improve the multiplexing gain over the OFDM system with fixed-position antenna (FPA) users, especially when robust channel coding is applied and the number of radio-frequency (RF) chains at each user is small.

cs.IT↗

arXiv2025

Higher Weil-Petersson volumes of the moduli space of super Riemann surfaces

Inspired by the theory of JT supergravity, Stanford-Witten derived a remarkable recursion formula of Weil-Petersson volumes of moduli space of super Riemann surfaces. It is the super version of the celebrated Mirzakhani's recursion formula. In this paper, we generalize Stanford-Witten's formula to include high degree kappa classes.

math.AG↗

arXiv2025

Analysis of a nonlinear free boundary problem modeling the radial growth of two-layer tumors

In this paper we study a nonlinear free boundary problem on the radial growth of a two-layer solid tumor with a quiescent core. The tumor surface and its inner interface separating the proliferating cells and the quiescent cells are both free boundaries. By deeply analyzing their relationship and employing the maximum principle, we show this problem is globally well-posed and prove the existence of a unique positive threshold $σ^*$ such that the problem admits a unique stationary solution with a quiescent core if and only if the externally supplied nutrient $\barσ> σ^*$. The stationary solution is globally asymptotically stable. The formation of the quiescent core and its interesting connection with the necrotic core are also given.

math.AP↗