arXiv subjects

Jie Zhang

Publications and source records attributed to Jie Zhang.

At least 361 records · Page 20Linked to original sources

arXiv2025

A Multi-Scale Graph Neural Process with Cross-Drug Co-Attention for Drug-Drug Interactions Prediction

Accurate prediction of drug-drug interactions (DDI) is crucial for medication safety and effective drug development. However, existing methods often struggle to capture structural information across different scales, from local functional groups to global molecular topology, and typically lack mechanisms to quantify prediction confidence. To address these limitations, we propose MPNP-DDI, a novel Multi-scale Graph Neural Process framework. The core of MPNP-DDI is a unique message-passing scheme that, by being iteratively applied, learns a hierarchy of graph representations at multiple scales. Crucially, a cross-drug co-attention mechanism then dynamically fuses these multi-scale representations to generate context-aware embeddings for interacting drug pairs, while an integrated neural process module provides principled uncertainty estimation. Extensive experiments demonstrate that MPNP-DDI significantly outperforms state-of-the-art baselines on benchmark datasets. By providing accurate, generalizable, and uncertainty-aware predictions built upon multi-scale structural features, MPNP-DDI represents a powerful computational tool for pharmacovigilance, polypharmacy risk assessment, and precision medicine.

cs.LG↗

arXiv2025

Taming Anomalies with Down-Up Sampling Networks: Group Center Preserving Reconstruction for 3D Anomaly Detection

Reconstruction-based methods have demonstrated very promising results for 3D anomaly detection. However, these methods face great challenges in handling high-precision point clouds due to the large scale and complex structure. In this study, a Down-Up Sampling Network (DUS-Net) is proposed to reconstruct high-precision point clouds for 3D anomaly detection by preserving the group center geometric structure. The DUS-Net first introduces a Noise Generation module to generate noisy patches, which facilitates the diversity of training data and strengthens the feature representation for reconstruction. Then, a Down-sampling Network (Down-Net) is developed to learn an anomaly-free center point cloud from patches with noise injection. Subsequently, an Up-sampling Network (Up-Net) is designed to reconstruct high-precision point clouds by fusing multi-scale up-sampling features. Our method leverages group centers for construction, enabling the preservation of geometric structure and providing a more precise point cloud. Extensive experiments demonstrate the effectiveness of our proposed method, achieving state-of-the-art (SOTA) performance with an Object-level AUROC of 79.9% and 79.5%, and a Point-level AUROC of 71.2% and 84.7% on the Real3D-AD and Anomaly-ShapeNet datasets, respectively.

cs.CV↗

arXiv2025

Digital Twin-Assisted Resilient Planning for mmWave IAB Networks via Graph Attention Networks

Digital Twin (DT) technology enables real-time monitoring and optimization of complex network infrastructures by creating accurate virtual replicas of physical systems. In millimeter-wave (mmWave) 5G/6G networks, the deployment of Integrated Access and Backhaul (IAB) nodes faces highly dynamic urban environments, necessitating intelligent DT-enabled optimization frameworks. Traditional IAB deployment optimization approaches struggle with the combinatorial complexity of jointly optimizing coverage, connectivity, and resilience, often leading to suboptimal solutions that are vulnerable to network disruptions. With this consideration, we propose a novel Graph Attention Network v2 (GATv2)-based reinforcement learning approach for resilient IAB deployment in urban mmWave networks. Specifically, we formulate the deployment problem as a Markov Decision Process (MDP) with explicit resilience constraints and employ edge-conditioned GATv2 to capture complex spatial dependencies between heterogeneous node types and dynamic connectivity patterns. The attention mechanism enables the model to focus on critical deployment locations to maximize coverage and ensure fault tolerance through redundant backhaul connections. To address the inherent vulnerability of mmWave links, we train the GATv2 policy using Proximal Policy Optimization (PPO) with a carefully designed balance between coverage, cost, and resilience. Comprehensive simulations across three urban scenarios demonstrate that our method achieves 98.5-98.7 percent coverage with 14.3-26.7 percent fewer nodes than baseline approaches, while maintaining 87.1 percent coverage retention under 30 percent link failures, representing 11.3-15.4 percent improvement in fault tolerance compared to state-of-the-art methods.

cs.NI↗

arXiv2025

XBOF: A Cost-Efficient CXL JBOF with Inter-SSD Compute Resource Sharing

Enterprise SSDs integrate numerous computing resources (e.g., ARM processor and onboard DRAM) to satisfy the ever-increasing performance requirements of I/O bursts. While these resources substantially elevate the monetary costs of SSDs, the sporadic nature of I/O bursts causes severe SSD resource underutilization in just a bunch of flash (JBOF) level. Tackling this challenge, we propose XBOF, a cost-efficient JBOF design, which only reserves moderate computing resources in SSDs at low monetary cost, while achieving demanded I/O performance through efficient inter-SSD resource sharing. Specifically, XBOF first disaggregates SSD architecture into multiple disjoint parts based on their functionality, enabling fine-grained SSD internal resource management. XBOF then employs a decentralized scheme to manage these disaggregated resources and harvests the computing resources of idle SSDs to assist busy SSDs in handling I/O bursts. This idea is facilitated by the cache-coherent capability of Compute eXpress Link (CXL), with which the busy SSDs can directly utilize the harvested computing resources to accelerate metadata processing. The evaluation results show that XBOF improves SSD resource utilization by 50.4% and saves 19.0% monetary costs with a negligible performance loss, compared to existing JBOF designs.

cs.OS↗

arXiv2025

Evaluating Cognitive-Behavioral Fixation via Multimodal User Viewing Patterns on Social Media

Digital social media platforms frequently contribute to cognitive-behavioral fixation, a phenomenon in which users exhibit sustained and repetitive engagement with narrow content domains. While cognitive-behavioral fixation has been extensively studied in psychology, methods for computationally detecting and evaluating such fixation remain underexplored. To address this gap, we propose a novel framework for assessing cognitive-behavioral fixation by analyzing users' multimodal social media engagement patterns. Specifically, we introduce a multimodal topic extraction module and a cognitive-behavioral fixation quantification module that collaboratively enable adaptive, hierarchical, and interpretable assessment of user behavior. Experiments on existing benchmarks and a newly curated multimodal dataset demonstrate the effectiveness of our approach, laying the groundwork for scalable computational analysis of cognitive fixation. All code in this project is publicly available for research purposes at https://github.com/Liskie/cognitive-fixation-evaluation.

cs.SI↗

arXiv2025

Estimating Cosmological Parameters and Reconstructing Hubble Constant with Artificial Neural Networks: A Test with covariance matrix and mock H(z)

In this work, we reconstruct the H(z) based on observational Hubble data with Artificial Neural Network, then estimate the cosmological parameters and the Hubble constant. The training data we used are covariance matrix and mock H(z), which are generated based on the real OHD data and Gaussian Process(GP). The use of the covariance matrix propagates the correlated uncertainties and improves training efficiency. Using the reconstructed H(z) data, we first determine the Hubble constant and compare it with CMB-based measurements. To constrain cosmological parameters, we sample on the reconstructed data and calculate the corresponding posterior distributions with Markov Chain Monte Carlo (MCMC). Through comprehensive statistical comparisons, we demonstrate that the parameter estimation using reconstructed samples achieves comparable statistical accuracy to the result derived from real OHD data.

astro-ph.CO↗

arXiv2025

TableZoomer: A Collaborative Agent Framework for Large-scale Table Question Answering

While large language models (LLMs) have shown promise in the table question answering (TQA) task through prompt engineering, they face challenges in industrial applications, including structural heterogeneity, difficulties in target data localization, and bottlenecks in complex reasoning. To address these limitations, this paper presents TableZoomer, a novel LLM-powered, programming-based agent framework. It introduces three key innovations: (1) replacing the original fully verbalized table with structured table schema to bridge the semantic gap and reduce computational complexity; (2) a query-aware table zooming mechanism that dynamically generates sub-table schema through column selection and entity linking, significantly improving target localization efficiency; and (3) a Program-of-Thoughts (PoT) strategy that transforms queries into executable code to mitigate numerical hallucination. Additionally, we integrate the reasoning workflow with the ReAct paradigm to enable iterative reasoning. Extensive experiments demonstrate that our framework maintains the usability advantages while substantially enhancing performance and scalability across tables of varying scales. When implemented with the Qwen3-8B-Instruct LLM, TableZoomer achieves accuracy improvements of 19.34% and 25% over conventional PoT methods on the large-scale DataBench dataset and the small-scale Fact Checking task of TableBench dataset, respectively.

cs.CL↗

arXiv2025

Topological signatures of collective dynamics and turbulent-like energy cascades in apolar active granular

Active matter refers to a broad class of non-equilibrium systems where energy is continuously injected at the level of individual ``particles". These systems exhibit emergent collective behaviors that have no direct thermal-equilibrium counterpart. Their scale ranges from micrometer-sized swarms of bacteria to meter-scale human crowds. In recent years, the role of topology and self-propelled topological defects in active systems has garnered significant attention, particularly in polar and nematic active matter. Building on these ideas, we investigate emergent collective dynamics in apolar active granular fluids. Using isotropic granular vibrators as a model experimental system of apolar active Ornstein-Uhlenbeck particles in a dry environment, we uncover a distinctive three-stage time evolution arising from the intricate interplay between activity and inelastic interactions. By analyzing the statistics, spatial correlations, and dynamics of vortex-like topological defects in the displacement vector field, we demonstrate their ability to describe this intrinsic collective motion. Furthermore, associated to these topological defects, we reveal the onset of a turbulent-like inverse energy cascade, where kinetic energy transfers across different length scales over time. As the system evolves, the power scaling of the energy transfer increases with the duration of observation. Our findings show that topological concepts can be extended to the nonequilibrium dynamics of apolar active matter, revealing a direct link between microscopic topological processes and emergent large-scale behaviors in active granular fluids that lack both a well-defined direction of motion and an intrinsic axis of orientation at the particle scale.

cond-mat.soft↗

arXiv2025

Beyond Frequency: The Role of Redundancy in Large Language Model Memorization

Memorization in large language models poses critical risks for privacy and fairness as these systems scale to billions of parameters. While previous studies established correlations between memorization and factors like token frequency and repetition patterns, we revealed distinct response patterns: frequency increases minimally impact memorized samples (e.g. 0.09) while substantially affecting non-memorized samples (e.g., 0.25), with consistency observed across model scales. Through counterfactual analysis by perturbing sample prefixes and quantifying perturbation strength through token positional changes, we demonstrate that redundancy correlates with memorization patterns. Our findings establish that: about 79% of memorized samples are low-redundancy, these low-redundancy samples exhibit 2-fold higher vulnerability than high-redundancy ones, and consequently memorized samples drop by 0.6 under perturbation while non-memorized samples drop by only 0.01, indicating that more redundant content becomes both more memorable and more fragile. These findings suggest potential redundancy-guided approaches for data preprocessing, thereby reducing privacy risks and mitigating bias to ensure fairness in model deployments.

cs.LG↗

arXiv2025

Evaluating Differentially Private Generation of Domain-Specific Text

Generative AI offers transformative potential for high-stakes domains such as healthcare and finance, yet privacy and regulatory barriers hinder the use of real-world data. To address this, differentially private synthetic data generation has emerged as a promising alternative. In this work, we introduce a unified benchmark to systematically evaluate the utility and fidelity of text datasets generated under formal Differential Privacy (DP) guarantees. Our benchmark addresses key challenges in domain-specific benchmarking, including choice of representative data and realistic privacy budgets, accounting for pre-training and a variety of evaluation metrics. We assess state-of-the-art privacy-preserving generation methods across five domain-specific datasets, revealing significant utility and fidelity degradation compared to real data, especially under strict privacy constraints. These findings underscore the limitations of current approaches, outline the need for advanced privacy-preserving data sharing methods and set a precedent regarding their evaluation in realistic scenarios.

cs.LG↗

arXiv2025

LABNet: A Lightweight Attentive Beamforming Network for Ad-hoc Multichannel Microphone Invariant Real-Time Speech Enhancement

Multichannel speech enhancement (SE) aims to restore clean speech from noisy measurements by leveraging spatiotemporal signal features. In ad-hoc array conditions, microphone invariance (MI) requires systems to handle different microphone numbers and array geometries. From a practical perspective, multichannel recordings inevitably increase the computational burden for edge-device applications, highlighting the necessity of lightweight and efficient deployments. In this work, we propose a lightweight attentive beamforming network (LABNet) to integrate MI in a low-complexity real-time SE system. We design a three-stage framework for efficient intra-channel modeling and inter-channel interaction. A cross-channel attention module is developed to aggregate features from each channel selectively. Experimental results demonstrate our LABNet achieves impressive performance with ultra-light resource overhead while maintaining the MI, indicating great potential for ad-hoc array processing. The code is available:https://github.com/Jokejiangv/LABNet.git

cs.SD↗

arXiv2025

FOCUS: Frequency-Optimized Conditioning of DiffUSion Models for mitigating catastrophic forgetting during Test-Time Adaptation

Test-time adaptation enables models to adapt to evolving domains. However, balancing the tradeoff between preserving knowledge and adapting to domain shifts remains challenging for model adaptation methods, since adapting to domain shifts can induce forgetting of task-relevant knowledge. To address this problem, we propose FOCUS, a novel frequency-based conditioning approach within a diffusion-driven input-adaptation framework. Utilising learned, spatially adaptive frequency priors, our approach conditions the reverse steps during diffusion-driven denoising to preserve task-relevant semantic information for dense prediction. FOCUS leverages a trained, lightweight, Y-shaped Frequency Prediction Network (Y-FPN) that disentangles high and low frequency information from noisy images. This minimizes the computational costs involved in implementing our approach in a diffusion-driven framework. We train Y-FPN with FrequencyMix, a novel data augmentation method that perturbs the images across diverse frequency bands, which improves the robustness of our approach to diverse corruptions. We demonstrate the effectiveness of FOCUS for semantic segmentation and monocular depth estimation across 15 corruption types and three datasets, achieving state-of-the-art averaged performance. In addition to improving standalone performance, FOCUS complements existing model adaptation methods since we can derive pseudo labels from FOCUS-denoised images for additional supervision. Even under limited, intermittent supervision with the pseudo labels derived from the FOCUS denoised images, we show that FOCUS mitigates catastrophic forgetting for recent model adaptation methods.

cs.CV↗

arXiv2025

Experimental Confirmation of Hyperuniformity of torque fluctuations in Frictional Matter

A recent purely theoretical prediction stated that in frictional granular packings, mechanical balance and material isotropy constrain the stress auto-correlation matrix to be fully determined by two spatially isotropic functions: the pressure and torque auto-correlations. Moreover, and unexpectedly, the torque fluctuations were predicted to be hyperuniform, a condition for the stress auto-correlation to decay as the elastic Green's function. In this Letter we present first experimental evidence for the hyperuniformity of the stress fluctuations. We propose that quite generally the variance of torque fluctuations in a d-dimensional ball of radius R will increase exactly like $R^{d-1}$, satisfying the definition of hyperuniformity.

cond-mat.soft↗

arXiv2025

SHALE: A Scalable Benchmark for Fine-grained Hallucination Evaluation in LVLMs

Despite rapid advances, Large Vision-Language Models (LVLMs) still suffer from hallucinations, i.e., generating content inconsistent with input or established world knowledge, which correspond to faithfulness and factuality hallucinations, respectively. Prior studies primarily evaluate faithfulness hallucination at a rather coarse level (e.g., object-level) and lack fine-grained analysis. Additionally, existing benchmarks often rely on costly manual curation or reused public datasets, raising concerns about scalability and data leakage. To address these limitations, we propose an automated data construction pipeline that produces scalable, controllable, and diverse evaluation data. We also design a hierarchical hallucination induction framework with input perturbations to simulate realistic noisy scenarios. Integrating these designs, we construct SHALE, a Scalable HALlucination Evaluation benchmark designed to assess both faithfulness and factuality hallucinations via a fine-grained hallucination categorization scheme. SHALE comprises over 30K image-instruction pairs spanning 12 representative visual perception aspects for faithfulness and 6 knowledge domains for factuality, considering both clean and noisy scenarios. Extensive experiments on over 20 mainstream LVLMs reveal significant factuality hallucinations and high sensitivity to semantic perturbations.

cs.CV↗

arXiv2025

Cowpox: Towards the Immunity of VLM-based Multi-Agent Systems

Vision Language Model (VLM)-based agents are stateful, autonomous entities capable of perceiving and interacting with their environments through vision and language. Multi-agent systems comprise specialized agents who collaborate to solve a (complex) task. A core security property is robustness, stating that the system should maintain its integrity under adversarial attacks. However, the design of existing multi-agent systems lacks the robustness consideration, as a successful exploit against one agent can spread and infect other agents to undermine the entire system's assurance. To address this, we propose a new defense approach, Cowpox, to provably enhance the robustness of multi-agent systems. It incorporates a distributed mechanism, which improves the recovery rate of agents by limiting the expected number of infections to other agents. The core idea is to generate and distribute a special cure sample that immunizes an agent against the attack before exposure and helps recover the already infected agents. We demonstrate the effectiveness of Cowpox empirically and provide theoretical robustness guarantees.

cs.MA↗

arXiv2025

Evolution of extreme nonlinear wave fields over strongly reflective plane beaches

The description of complex wave processes, in addition to the shoaling problem, is often cumbersome even for the evolution of regular waves. For reflection under the regime of wave breaking, the surf similarity is generally accepted as the leading parameter controlling the reflection rates and types of breakers. Although little is known about the effect of reflection rates on the formation of extreme nonlinear waves, some debate has arisen regarding whether high reflection rates enhance the nonlinearity at the tail of the wave height distribution. In this work, we provide theoretical evidence that at very steep beaches of smooth composition, the reflection rate near unity will tend to stabilize the excess kurtosis otherwise generated by shoaling and controlled in magnitude by the bottom slope magnitude. We further verified this result through fully nonlinear numerical simulations, reaching a good agreement.

nlin.PS↗

arXiv2025

TAPAS: Fast and Automatic Derivation of Tensor Parallel Strategies for Large Neural Networks

Tensor parallelism is an essential technique for distributed training of large neural networks. However, automatically determining an optimal tensor parallel strategy is challenging due to the gigantic search space, which grows exponentially with model size and tensor dimension. This prohibits the adoption of auto-parallel systems on larger models. We observe that neural networks usually contain repeated substructures, and build an automatic parallelism framework named TAPAS that eliminates redundant search efforts. TAPAS employs a divide-and-conquer approach that efficiently folds the search space by identifying those unique substructures. As a result, it runs at sub-linear complexity concerning the model size, making it a scalable solution for training large-scale networks. Our evaluations demonstrate that TAPAS outperforms the state-of-the-art automatic parallelism frameworks by up to $160\times$ in search speed on a wide range of models, and the performance of derived strategies is competitive or even better compared with the expert-engineered Megatron-LM library.

cs.LG↗

arXiv2025

PoseGuard: Pose-Guided Generation with Safety Guardrails

Pose-guided video generation has become a powerful tool in creative industries, exemplified by frameworks like Animate Anyone. However, conditioning generation on specific poses introduces serious risks, such as impersonation, privacy violations, and NSFW content creation. To address these challenges, we propose $\textbf{PoseGuard}$, a safety alignment framework for pose-guided generation. PoseGuard is designed to suppress unsafe generations by degrading output quality when encountering malicious poses, while maintaining high-fidelity outputs for benign inputs. We categorize unsafe poses into three representative types: discriminatory gestures such as kneeling or offensive salutes, sexually suggestive poses that lead to NSFW content, and poses imitating copyrighted celebrity movements. PoseGuard employs a dual-objective training strategy combining generation fidelity with safety alignment, and uses LoRA-based fine-tuning for efficient, parameter-light updates. To ensure adaptability to evolving threats, PoseGuard supports pose-specific LoRA fusion, enabling flexible and modular updates when new unsafe poses are identified. We further demonstrate the generalizability of PoseGuard to facial landmark-guided generation. Extensive experiments validate that PoseGuard effectively blocks unsafe generations, maintains generation quality for benign inputs, and remains robust against slight pose variations.

cs.CR↗