arXiv subjects

Rui Zhang

Publications and source records attributed to Rui Zhang.

At least 325 records · Page 18Linked to original sources

arXiv2025

Large-scale Multi-sequence Pretraining for Generalizable MRI Analysis in Versatile Clinical Applications

Multi-sequence Magnetic Resonance Imaging (MRI) offers remarkable versatility, enabling the distinct visualization of different tissue types. Nevertheless, the inherent heterogeneity among MRI sequences poses significant challenges to the generalization capability of deep learning models. These challenges undermine model performance when faced with varying acquisition parameters, thereby severely restricting their clinical utility. In this study, we present PRISM, a foundation model PRe-trained with large-scale multI-Sequence MRI. We collected a total of 64 datasets from both public and private sources, encompassing a wide range of whole-body anatomical structures, with scans spanning diverse MRI sequences. Among them, 336,476 volumetric MRI scans from 34 datasets (8 public and 26 private) were curated to construct the largest multi-organ multi-sequence MRI pretraining corpus to date. We propose a novel pretraining paradigm that disentangles anatomically invariant features from sequence-specific variations in MRI, while preserving high-level semantic representations. We established a benchmark comprising 44 downstream tasks, including disease diagnosis, image segmentation, registration, progression prediction, and report generation. These tasks were evaluated on 32 public datasets and 5 private cohorts. PRISM consistently outperformed both non-pretrained models and existing foundation models, achieving first-rank results in 39 out of 44 downstream benchmarks with statistical significance improvements. These results underscore its ability to learn robust and generalizable representations across unseen data acquired under diverse MRI protocols. PRISM provides a scalable framework for multi-sequence MRI analysis, thereby enhancing the translational potential of AI in radiology. It delivers consistent performance across diverse imaging protocols, reinforcing its clinical applicability.

eess.IV↗

arXiv2025

An Outlook on the Opportunities and Challenges of Multi-Agent AI Systems

A multi-agent AI system (MAS) is composed of multiple autonomous agents that interact, exchange information, and make decisions based on internal generative models. Recent advances in large language models and tool-using agents have made MAS increasingly practical in areas like scientific discovery and collaborative automation. However, key questions remain: When are MAS more effective than single-agent systems? What new safety risks arise from agent interactions? And how should we evaluate their reliability and structure? This paper outlines a formal framework for analyzing MAS, focusing on two core aspects: effectiveness and safety. We explore whether MAS truly improve robustness, adaptability, and performance, or merely repackage known techniques like ensemble learning. We also study how inter-agent dynamics may amplify or suppress system vulnerabilities. While MAS are relatively new to the signal processing community, we envision them as a powerful abstraction that extends classical tools like distributed estimation and sensor fusion to higher-level, policy-driven inference. Through experiments on data science automation, we highlight the potential of MAS to reshape how signal processing systems are designed and trusted.

cs.MA↗

arXiv2025

Comprehensive Structural Characterization of Charged Polymers Involved in Moisture-Driven Direct Air Capture

The rise in atmospheric carbon dioxide (CO2) levels has led to urgent calls for effective carbon capture methods, with direct air capture (DAC) emerging as a promising solution. This study focuses on the structural characterization of commercially available alkaline anion-exchange membrane (AEM) polymers, Fumasep FAA-3 and IRA 900, for use in low-energy, moisture-driven DAC applications. A combination of X-ray diffraction, small and wide-angle X-ray scattering (SAXS/WAXS), atomic force microscopy (AFM), focused ion beam-scanning electron microscopy (FIB-SEM), and transmission electron microscopy (TEM) were employed to explore the structural features of these materials. X-ray scattering analysis revealed molecular ordering and large-scale structural organization in both materials, while humidity-induced changes highlighted the impact of moisture on structural properties. AFM surface analysis further indicated the presence of clustering, porosity, and swelling, which were corroborated by FIB-SEM and TEM imaging. These structural insights offer a deeper understanding of the behavior of AEM-DAC materials during CO2 capture and release, emphasizing the role of moisture in these processes. This work lays the foundation for the development of more energy-efficient DAC polymers, paving the way for improved CO2 capture technologies.

cond-mat.mtrl-sci↗

arXiv2025

Combining Machine Learning Defenses without Conflicts

Machine learning (ML) defenses protect against various risks to security, privacy, and fairness. Real-life models need simultaneous protection against multiple different risks which necessitates combining multiple defenses. But combining defenses with conflicting interactions in an ML model can be ineffective, incurring a significant drop in the effectiveness of one or more defenses being combined. Practitioners need a way to determine if a given combination can be effective. Experimentally identifying effective combinations can be time-consuming and expensive, particularly when multiple defenses need to be combined. We need an inexpensive, easy-to-use combination technique to identify effective combinations. Ideally, a combination technique should be (a) accurate (correctly identifies whether a combination is effective or not), (b) scalable (allows combining multiple defenses), (c) non-invasive (requires no change to the defenses being combined), and (d) general (is applicable to different types of defenses). Prior works have identified several ad-hoc techniques but none satisfy all the requirements above. We propose a principled combination technique, Def\Con, to identify effective defense combinations. Def\Con meets all requirements, achieving 90% accuracy on eight combinations explored in prior work and 81% in 30 previously unexplored combinations that we empirically evaluate in this paper.

cs.CR↗

arXiv2025

Request-Only Optimization for Recommendation Systems

Deep Learning Recommendation Models (DLRMs) represent one of the largest machine learning applications on the planet. Industry-scale DLRMs are trained with petabytes of recommendation data to serve billions of users every day. To utilize the rich user signals in the long user history, DLRMs have been scaled up to unprecedented complexity, up to trillions of floating-point operations (TFLOPs) per example. This scale, coupled with the huge amount of training data, necessitates new storage and training algorithms to efficiently improve the quality of these complex recommendation systems. In this paper, we present a Request-Only Optimizations (ROO) training and modeling paradigm. ROO simultaneously improves the storage and training efficiency as well as the model quality of recommendation systems. We holistically approach this challenge through co-designing data (i.e., request-only data), infrastructure (i.e., request-only based data processing pipeline), and model architecture (i.e., request-only neural architectures). Our ROO training and modeling paradigm treats a user request as a unit of the training data. Compared with the established practice of treating a user impression as a unit, our new design achieves native feature deduplication in data logging, consequently saving data storage. Second, by de-duplicating computations and communications across multiple impressions in a request, this new paradigm enables highly scaled-up neural network architectures to better capture user interest signals, such as Generative Recommenders (GRs) and other request-only friendly architectures.

cs.IR↗

arXiv2025

Data-Efficient Biomedical In-Context Learning: A Diversity-Enhanced Submodular Perspective

Recent progress in large language models (LLMs) has leveraged their in-context learning (ICL) abilities to enable quick adaptation to unseen biomedical NLP tasks. By incorporating only a few input-output examples into prompts, LLMs can rapidly perform these new tasks. While the impact of these demonstrations on LLM performance has been extensively studied, most existing approaches prioritize representativeness over diversity when selecting examples from large corpora. To address this gap, we propose Dual-Div, a diversity-enhanced data-efficient framework for demonstration selection in biomedical ICL. Dual-Div employs a two-stage retrieval and ranking process: First, it identifies a limited set of candidate examples from a corpus by optimizing both representativeness and diversity (with optional annotation for unlabeled data). Second, it ranks these candidates against test queries to select the most relevant and non-redundant demonstrations. Evaluated on three biomedical NLP tasks (named entity recognition (NER), relation extraction (RE), and text classification (TC)) using LLaMA 3.1 and Qwen 2.5 for inference, along with three retrievers (BGE-Large, BMRetriever, MedCPT), Dual-Div consistently outperforms baselines-achieving up to 5% higher macro-F1 scores-while demonstrating robustness to prompt permutations and class imbalance. Our findings establish that diversity in initial retrieval is more critical than ranking-stage optimization, and limiting demonstrations to 3-5 examples maximizes performance efficiency.

cs.CL↗

arXiv2025

Quasi-Static IRS: 3D Shaped Beamforming for Area Coverage Enhancement

Intelligent reflecting surface (IRS) is a promising paradigm to reconfigure the wireless environment for enhanced communication coverage and quality. However, to compensate for the double pathloss effect, massive IRS elements are required, raising concerns on the scalability of cost and complexity. This paper introduces a new architecture of quasi-static IRS (QS-IRS), which tunes element phases via mechanical adjustment or manually re-arranging the array topology. QS-IRS relies on massive production/assembly of purely passive elements only, and thus is suitable for ultra low-cost and large-scale deployment to enhance long-term coverage. To achieve this end, an IRS-aided area coverage problem is formulated, which explicitly considers the element radiation pattern (ERP), with the newly introduced shape masks for the mainlobe, and the sidelobe constraints to reduce energy leakage. An alternating optimization (AO) algorithm based on the difference-of-convex (DC) and successive convex approximation (SCA) procedure is proposed, which achieves shaped beamforming with power gains close to that of the joint optimization algorithm, but with significantly reduced computational complexity.

cs.IT↗

arXiv2025

OmniFluids: Physics Pre-trained Modeling of Fluid Dynamics

Computational fluid dynamics (CFD) drives progress in numerous scientific and engineering fields, yet high-fidelity simulations remain computationally prohibitive. While machine learning approaches offer computing acceleration, they typically specialize in single physical systems or require extensive training data, hindering their applicability in highly nonlinear and 3D flow scenarios. To overcome these limitations, we propose OmniFluids, a pure physics pre-trained model that captures fundamental fluid dynamics laws and adapts efficiently to diverse downstream tasks with minimal data. We develop a training framework combining physics-only pre-training, coarse-grid operator distillation, and few-shot fine-tuning. This enables OmniFluids to retain broad physics knowledge while delivering fast and accurate predictions. Architecturally, OmniFluids integrates a mixture of operators, a multi-frame decoder, and factorized Fourier layers, seamlessly incorporating physics-based supervision while allowing efficient and scalable modeling of diverse tasks. Extensive tests on a broad range of 2D and 3D benchmarks show that OmniFluids outperforms state-of-the-art AI-driven methods in terms of flow field prediction and turbulence statistics. It delivers 10--100$\times$ speedups over traditional solvers while maintaining a comparable accuracy and accurately identifies unknown physical parameters from sparse, noisy data. This work demonstrates the potential of training a unified CFD solver exclusively from physics knowledge, offering a new approach for efficient and generalizable modeling across complex fluid systems.

physics.flu-dyn↗

arXiv2025

HERGC: Heterogeneous Experts Representation and Generative Completion for Multimodal Knowledge Graphs

Multimodal knowledge graphs (MMKGs) enrich traditional knowledge graphs (KGs) by incorporating diverse modalities such as images and text. multimodal knowledge graph completion (MMKGC) seeks to exploit these heterogeneous signals to infer missing facts, thereby mitigating the intrinsic incompleteness of MMKGs. Existing MMKGC methods typically leverage only the information contained in the MMKGs under the closed-world assumption and adopt discriminative training objectives, which limits their reasoning capacity during completion. Recent large language models (LLMs), empowered by massive parameter scales and pretraining on vast corpora, have demonstrated strong reasoning abilities across various tasks. However, their potential in MMKGC remains largely unexplored. To bridge this gap, we propose HERGC, a flexible Heterogeneous Experts Representation and Generative Completion framework for MMKGs. HERGC first deploys a Heterogeneous Experts Representation Retriever that enriches and fuses multimodal information and retrieves a compact candidate set for each incomplete triple. It then uses a Generative LLM Predictor, implemented via either in-context learning or lightweight fine-tuning, to accurately identify the correct answer from these candidates. Extensive experiments on three standard MMKG benchmarks demonstrate HERGC's effectiveness and robustness, achieving superior performance over existing methods.

cs.CL↗

arXiv2025

3D-CDRGP: Towards Cross-Device Robotic Grasping Policy in 3D Open World

Given the diversity of devices and the product upgrades, cross-device research has become an urgent issue that needs to be tackled. To this end, we pioneer in probing the cross-device (cameras & robotics) grasping policy in the 3D open world. Specifically, we construct two real-world grasping setups, employing robotic arms and cameras from completely different manufacturers. To minimize domain differences in point clouds from diverse cameras, we adopt clustering methods to generate 3D object proposals. However, existing clustering methods are limited to closed-set scenarios, which confines the robotic graspable object categories and ossifies the deployment scenarios. To extend these methods to open-world settings, we introduce the SSGC-Seg module that enables category-agnostic 3D object detection. The proposed module transforms the original multi-class semantic information into binary semantic cues-foreground and background by analyzing the SoftMax value of each point, and then clusters the foreground points based on geometric information to form initial object proposals. Furthermore, ScoreNet‡ is designed to score each detection result, and the robotic arm prioritizes grasping the object with the highest confidence score. Experiments on two different types of setups highlight the effectiveness and robustness of our policy for cross-device robotics grasping research. Our code is provided in the supplementary and will be released upon acceptance.

cs.RO↗

arXiv2025

RoboGSim: A Real2Sim2Real Robotic Gaussian Splatting Simulator

Efficient acquisition of real-world embodied data has been increasingly critical. However, large-scale demonstrations captured by remote operation tend to take extremely high costs and fail to scale up the data size in an efficient manner. Sampling the episodes under a simulated environment is a promising way for large-scale collection while existing simulators fail to high-fidelity modeling on texture and physics. To address these limitations, we introduce the RoboGSim, a real2sim2real robotic simulator, powered by 3D Gaussian Splatting and the physics engine. RoboGSim mainly includes four parts: Gaussian Reconstructor, Digital Twins Builder, Scene Composer, and Interactive Engine. It can synthesize the simulated data with novel views, objects, trajectories, and scenes. RoboGSim also provides an online, reproducible, and safe evaluation for different manipulation policies. The real2sim and sim2real transfer experiments show a high consistency in the texture and physics. We compared the test results of RoboGSim data and real robot data on both RoboGSim and real robot platforms. The experimental results show that the RoboGSim data model can achieve zero-shot performance on the real robot, with results comparable to real robot data. Additionally, in experiments with novel perspectives and novel scenes, the RoboGSim data model performed even better on the real robot than the real robot data model. This not only helps reduce the sim2real gap but also addresses the limitations of real robot data collection, such as its single-source and high cost. We hope RoboGSim serves as a closed-loop simulator for fair comparison on policy learning. More information can be found on our project page https://robogsim.github.io/.

cs.RO↗

arXiv2025

Geminio: Language-Guided Gradient Inversion Attacks in Federated Learning

Foundation models that bridge vision and language have made significant progress. While they have inspired many life-enriching applications, their potential for abuse in creating new threats remains largely unexplored. In this paper, we reveal that vision-language models (VLMs) can be weaponized to enhance gradient inversion attacks (GIAs) in federated learning (FL), where an FL server attempts to reconstruct private data samples from gradients shared by victim clients. Despite recent advances, existing GIAs struggle to reconstruct high-resolution images when the victim has a large local data batch. One promising direction is to focus reconstruction on valuable samples rather than the entire batch, but current methods lack the flexibility to target specific data of interest. To address this gap, we propose Geminio, the first approach to transform GIAs into semantically meaningful, targeted attacks. It enables a brand new privacy attack experience: attackers can describe, in natural language, the data they consider valuable, and Geminio will prioritize reconstruction to focus on those high-value samples. This is achieved by leveraging a pretrained VLM to guide the optimization of a malicious global model that, when shared with and optimized by a victim, retains only gradients of samples that match the attacker-specified query. Geminio can be launched at any FL round and has no impact on normal training (i.e., the FL server can steal clients' data while still producing a high-utility ML model as in benign scenarios). Extensive experiments demonstrate its effectiveness in pinpointing and reconstructing targeted samples, with high success rates across complex datasets and large batch sizes with resilience against defenses.

cs.LG↗

arXiv2025

Towed Movable Antenna (ToMA) Array for Ultra Secure Airborne Communications

This paper proposes a novel towed movable antenna (ToMA) array architecture to enhance the physical layer security of airborne communication systems. Unlike conventional onboard arrays with fixed-position antennas (FPAs), the ToMA array employs multiple subarrays mounted on flexible cables and towed by distributed drones, enabling agile deployment in three-dimensional (3D) space surrounding the central aircraft. This design significantly enlarges the effective array aperture and allows dynamic geometry reconfiguration, offering superior spatial resolution and beamforming flexibility. We consider a secure transmission scenario where an airborne transmitter communicates with multiple legitimate users in the presence of potential eavesdroppers. To ensure security, zero-forcing beamforming is employed to nullify signal leakage toward eavesdroppers. Based on the statistical distributions of locations of users and eavesdroppers, the antenna position vector (APV) of the ToMA array is optimized to maximize the users' ergodic achievable rate. Analytical results for the case of a single user and a single eavesdropper reveal the optimal APV structure that minimizes their channel correlation. For the general multiuser scenario, we develop a low-complexity alternating optimization algorithm by leveraging Riemannian manifold optimization. Simulation results confirm that the proposed ToMA array achieves significant performance gains over conventional onboard FPA arrays, especially in scenarios where eavesdroppers are closely located to users under line-of-sight (LoS)-dominant channels.

cs.IT↗

arXiv2025

Near-Field Communication with Massive Movable Antennas: A Functional Perspective

The advent of massive multiple-input multiple-output (MIMO) technology has provided new opportunities for capacity improvement via strategic antenna deployment, especially when the near-field effect is pronounced due to antenna proliferation. In this paper, we investigate the optimal antenna placement for maximizing the achievable rate of a point-to-point near-field channel, where the transmitter is deployed with massive movable antennas. First, we propose a novel design framework to explore the relationship between antenna positions and achievable data rate. By introducing the continuous antenna position function (APF) and antenna density function (ADF), we reformulate the antenna position design problem from the discrete to the continuous domain, which maximizes the achievable rate functional with respect to ADF. Leveraging functional analysis and variational methods, we derive the optimal ADF condition and propose a gradient-based algorithm for numerical solutions under general channel conditions. Furthermore, for the near-field line-of-sight (LoS) scenario, we present a closed-form solution for the optimal ADF, revealing the critical role of edge antenna density in enhancing the achievable rate. Finally, we propose a flexible antenna array-based deployment method that ensures practical implementation while mitigating mutual coupling issues. Simulation results demonstrate the effectiveness of the proposed framework, with uniform circular arrays emerging as a promising geometry for balancing performance and deployment feasibility in near-field communications.

cs.IT↗

arXiv2025

Dimension-reduced Reconstruction Map Learning for Parameter Estimation in Likelihood-Free Inference Problems

Many application areas rely on models that can be readily simulated but lack a closed-form likelihood, or an accurate approximation under arbitrary parameter values. Existing parameter estimation approaches in this setting are generally approximate. Recent work on using neural network models to reconstruct the mapping from the data space to the parameters from a set of synthetic parameter-data pairs suffers from the curse of dimensionality, resulting in inaccurate estimation as the data size grows. We propose a dimension-reduced approach to likelihood-free estimation which combines the ideas of reconstruction map estimation with dimension-reduction approaches based on subject-specific knowledge. We examine the properties of reconstruction map estimation with and without dimension reduction and explore the trade-off between approximation error due to information loss from reducing the data dimension and approximation error. Numerical examples show that the proposed approach compares favorably with reconstruction map estimation, approximate Bayesian computation, and synthetic likelihood estimation.

stat.ME↗

arXiv2025

Resummation for Lattice QCD Calculation of Generalized Parton Distributions at Nonzero Skewness

Large-momentum effective theory (LaMET) provides an approach to directly calculate the $x$-dependence of generalized parton distributions (GPDs) on a Euclidean lattice through power expansion and a perturbative matching. When a parton's momentum becomes soft, the corresponding logarithms in the matching kernel become non-negligible at higher orders of perturbation theory, which requires a resummation. But the resummation for the off-forward matrix elements at nonzero skewness $ξ$ is difficult due to their multi-scale nature. In this work, we demonstrate that these logarithms are important only in the threshold limit, and derive the threshold factorization formula for the quasi-GPDs in LaMET. We then propose an approach to resum all the large logarithms based on the threshold factorization, which is implemented on a GPD model. We demonstrate that the LaMET prediction is reliable for $[-1+x_0,-ξ-x_0]\cup[-ξ+x_0,ξ-x_0]\cup[ξ+x_0,1-x_0]$, where $x_0$ is a cutoff depending on hard parton momenta. Through our numerical tests with the GPD model, we demonstrate that our method is self-consistent and that the inverse matching does not spread the nonperturbative effects or power corrections to the perturbatively calculable regions.

hep-ph↗

arXiv2025

Low-Complexity 6DMA Rotation and Position Optimization Based on Statistical Channel Information

The six-dimensional movable antenna (6DMA) is a promising technology to fully exploit spatial variation in wireless channels by allowing flexible adjustment of three-dimensional (3D) positions and rotations of antennas at the transceiver. In this paper, we consider a 6DMA-equipped base station (BS) and aim to maximize the average sum logarithmic rate of all users served by the BS by jointly designing 6DMA surface positions and rotations based on statistical channel information (SCI). Different from prior works on 6DMA design which use alternating optimization to iteratively update surface positions and rotations, we propose a new sequential optimization method that first determines the optimal rotations and then identifies feasible positions to realize these rotations under practical antenna placement constraints. Simulation results show that our proposed optimization scheme significantly reduces the computational complexity of conventional alternating optimization (AO), while achieving communication performance comparable to the AO-based approach and superior to existing fixed-position/rotation antenna arrays.

cs.IT↗

arXiv2025

Rotational Metric: A Solution to Einstein's Clock-Rate Problem and Its Magnetospheric Applications

The rotational metric provides an exact solution to Einstein's clock-rate problem in curved spacetime, specifically, whether time flows more slowly at the equator of a compact object such as a neutron star than at its poles. It features a curvature singularity, an event horizon, a potentially evolving ergosphere, a rigidly-rotating normal space, and two stationary limit surfaces. Although derived from the Schwarzschild metric through rotational transformations, it includes an additional ergosphere. Given the equivalence of inertia and gravity, this demonstrates how non-inertial transformations, such as rotational transformations, can introduce new spacetime structures into a gravitational system. In particular, the additional physical degrees of freedom carried by rotational transformations are `eaten' by the gravitational system to form an additional ergosphere. Furthermore, the rotational metric effectively models a rigidly-rotating gravitational system and is applicable for describing rotationally-induced gravitational effects in various rotating magnetospheres.

gr-qc↗