arXiv subjects

Di Wu

Publications and source records attributed to Di Wu.

At least 199 records · Page 11Linked to original sources

arXiv2025

Chemical vapor deposition growth of continuous monolayer antiferromagnetic CrOCl films

The discovery of two-dimensional magnetic materials has provided an ideal platform for exploring physical phenomena in the two-dimensional limit. However, intrinsic two-dimensional antiferromagnetic materials have been rarely reported, limiting systematic studies of their electronic properties. The discovery of novel intrinsic two-dimensional antiferromagnets and the development of robust synthesis strategies, therefore, remain significant challenges. Here, we report the chemical vapor deposition synthesis of CrOCl monolayer films and nanosheets that exhibit excellent air stability. The CrOCl morphology is tunable, ranging from two-dimensional nanosheets to three-dimensional flower-like structures, with lateral sizes ranging from several microns to continuous monolayer films. Structural characterization confirms the materials composition and high crystalline quality. Furthermore, magnetic measurements, supported by theoretical calculations, reveal a Néel temperature for CrOCl of ~14 K. This work provides a reliable route for preparing two-dimensional antiferromagnetic materials.

cond-mat.mes-hall↗

arXiv2025

LEMUR: Large scale End-to-end MUltimodal Recommendation

Traditional ID-based recommender systems often struggle with cold-start and generalization challenges. Multimodal recommendation systems, which leverage textual and visual data, offer a promising solution to mitigate these issues. However, existing industrial approaches typically adopt a two-stage training paradigm: first pretraining a multimodal model, then applying its frozen representations to train the recommendation model. This decoupled framework suffers from misalignment between multimodal learning and recommendation objectives, as well as an inability to adapt dynamically to new data. To address these limitations, we propose LEMUR, the first large-scale multimodal recommender system trained end-to-end from raw data. By jointly optimizing both the multimodal and recommendation components, LEMUR ensures tighter alignment with downstream objectives while enabling real-time parameter updates. Constructing multimodal sequential representations from user history often entails prohibitively high computational costs. To alleviate this bottleneck, we propose a novel memory bank mechanism that incrementally accumulates historical multimodal representations throughout the training process. After one month of deployment in Douyin Search, LEMUR has led to a 0.843% reduction in query change rate decay and a 0.81% improvement in QAUC. Additionally, LEMUR has shown significant gains across key offline metrics for Douyin Advertisement. Our results validate the superiority of end-to-end multimodal recommendation in real-world industrial scenarios.

cs.IR↗

arXiv2025

$\widetilde{W}^{1+}$ subclass: Extending the topological classification of black hole thermodynamics

In this paper, we identify a novel topological subclass, dubbed $\widetilde{W}^{1+}$, in the thermodynamics of higher odd-dimensional, multiply rotating Kerr-AdS black holes. This discovery extends the established topological classification beyond the five classes and two subclasses previously known. The $\widetilde{W}^{1+}$ subclass exhibits a unique and previously unreported stability profile: it admits a thermodynamically stable small black hole state in the low-temperature limit, while in the high-temperature limit, the phase space simultaneously contains one stable large black hole, one stable small black hole, and one unstable small black hole state. Our analysis, which treats black hole solutions as topological defects, reveals a richer landscape of black hole thermodynamics than previously understood and necessitates an expansion of the topological classification scheme to accommodate this new phenomenology.

hep-th↗

arXiv2025

Towards Effective Federated Graph Foundation Model via Mitigating Knowledge Entanglement

Recent advances in graph machine learning have shifted to data-centric paradigms, driven by two emerging fields: (1) Federated graph learning (FGL) enables multi-client collaboration but faces challenges from data and task heterogeneity, limiting its practicality; (2) Graph foundation models (GFM) offer strong domain generalization but are usually trained on single machines, missing out on cross-silo data and resources. These paradigms are complementary, and their integration brings notable benefits. Motivated by this, we propose FedGFM, a novel decentralized GFM training paradigm. However, a key challenge is knowledge entanglement, where multi-domain knowledge merges into indistinguishable representations, hindering downstream adaptation. To address this, we present FedGFM+, an enhanced framework with two core modules to reduce knowledge entanglement: (1) AncDAI: A global anchor-based domain-aware initialization strategy. Before pre-training, each client encodes its local graph into domain-specific prototypes that serve as semantic anchors. Synthetic embeddings around these anchors initialize the global model. We theoretically prove these prototypes are distinguishable across domains, providing a strong inductive bias to disentangle domain-specific knowledge. (2) AdaDPP: A local adaptive domain-sensitive prompt pool. Each client learns a lightweight graph prompt capturing domain semantics during pre-training. During fine-tuning, prompts from all clients form a pool from which the GFM selects relevant prompts to augment target graph attributes, improving downstream adaptation. FedGFM+ is evaluated on 8 diverse benchmarks across multiple domains and tasks, outperforming 20 baselines from supervised learning, FGL, and federated GFM variants.

cs.LG↗

arXiv2025

Cross-Modal Unlearning via Influential Neuron Path Editing in Multimodal Large Language Models

Multimodal Large Language Models (MLLMs) extend foundation models to real-world applications by integrating inputs such as text and vision. However, their broad knowledge capacity raises growing concerns about privacy leakage, toxicity mitigation, and intellectual property violations. Machine Unlearning (MU) offers a practical solution by selectively forgetting targeted knowledge while preserving overall model utility. When applied to MLLMs, existing neuron-editing-based MU approaches face two fundamental challenges: (1) forgetting becomes inconsistent across modalities because existing point-wise attribution methods fail to capture the structured, layer-by-layer information flow that connects different modalities; and (2) general knowledge performance declines when sensitive neurons that also support important reasoning paths are pruned, as this disrupts the model's ability to generalize. To alleviate these limitations, we propose a multimodal influential neuron path editor (MIP-Editor) for MU. Our approach introduces modality-specific attribution scores to identify influential neuron paths responsible for encoding forget-set knowledge and applies influential-path-aware neuron-editing via representation misdirection. This strategy also enables effective and coordinated forgetting across modalities while preserving the model's general capabilities. Experimental results demonstrate that MIP-Editor achieves a superior unlearning performance on multimodal tasks, with a maximum forgetting rate of 87.75% and up to 54.26% improvement in general knowledge retention. On textual tasks, MIP-Editor achieves up to 80.65% forgetting and preserves 77.9% of general performance. Codes are available at https://github.com/PreckLi/MIP-Editor.

cs.LG↗

arXiv2025

Rethinking Crystal Symmetry Prediction: A Decoupled Perspective

Efficiently and accurately determining the symmetry is a crucial step in the structural analysis of crystalline materials. Existing methods usually mindlessly apply deep learning models while ignoring the underlying chemical rules. More importantly, experiments show that they face a serious sub-property confusion SPC problem. To address the above challenges, from a decoupled perspective, we introduce the XRDecoupler framework, a problem-solving arsenal specifically designed to tackle the SPC problem. Imitating the thinking process of chemists, we innovatively incorporate multidimensional crystal symmetry information as superclass guidance to ensure that the model's prediction process aligns with chemical intuition. We further design a hierarchical PXRD pattern learning model and a multi-objective optimization approach to achieve high-quality representation and balanced optimization. Comprehensive evaluations on three mainstream databases (e.g., CCDC, CoREMOF, and InorganicData) demonstrate that XRDecoupler excels in performance, interpretability, and generalization.

cs.LG↗

arXiv2025

NVIDIA Nemotron Nano V2 VL

We introduce Nemotron Nano V2 VL, the latest model of the Nemotron vision-language series designed for strong real-world document understanding, long video comprehension, and reasoning tasks. Nemotron Nano V2 VL delivers significant improvements over our previous model, Llama-3.1-Nemotron-Nano-VL-8B, across all vision and text domains through major enhancements in model architecture, datasets, and training recipes. Nemotron Nano V2 VL builds on Nemotron Nano V2, a hybrid Mamba-Transformer LLM, and innovative token reduction techniques to achieve higher inference throughput in long document and video scenarios. We are releasing model checkpoints in BF16, FP8, and FP4 formats and sharing large parts of our datasets, recipes and training code.

cs.LG↗

arXiv2025

Structural Priors and Modular Adapters in the Composable Fine-Tuning Algorithm of Large-Scale Models

This paper proposes a composable fine-tuning method that integrates graph structural priors with modular adapters to address the high computational cost and structural instability faced by large-scale pre-trained models in multi-task adaptation. The method introduces a relation matrix to model dependencies among tasks, explicitly encoding correlations between nodes and paths into graph structural priors, which provide unified structural constraints for adapter weight allocation and path selection. Modular adapters are embedded into different layers through low-rank mapping and a pluggable mechanism, enabling efficient cross-task composition and reuse under prior guidance. This mechanism not only improves parameter efficiency and training stability but also alleviates path conflicts and redundant computation in multi-task scenarios. Furthermore, experiments on hyperparameter sensitivity, environmental sensitivity, and data sensitivity are conducted to systematically analyze key factors such as routing temperature, gating thresholds, and relation matrix regularization strength, verifying the consistency and superior performance of the method under structural constraints. The results demonstrate that the proposed framework significantly enhances task prediction accuracy, adapter weight allocation precision, and overall computational efficiency while maintaining model lightweight design, highlighting the synergistic advantages of graph priors and modular mechanisms in composable fine-tuning.

cs.LG↗

arXiv2025

Dynamic Topic Evolution with Temporal Decay and Attention in Large Language Models

This paper proposes a modeling framework for dynamic topic evolution based on temporal large language models. The method first uses a large language model to obtain contextual embeddings of text and then introduces a temporal decay function and an attention mechanism. These components allow the model to adjust the importance of semantic units according to time intervals and capture topic variations across different periods. The temporal representations are then mapped into a latent topic space, where a state transition matrix is applied to describe the dynamic evolution of topics. A joint optimization objective constrains both semantic modeling and temporal consistency, ensuring diversity and smoothness in topic generation. The design emphasizes the unified modeling of semantic representation and temporal evolution, which improves topic coherence and diversity while enhancing stability and interpretability over time. Experiments on real-world corpora show that the framework effectively captures the generation, expansion, and decline of topics and outperforms existing models across multiple metrics. Overall, the proposed method provides a systematic solution for understanding dynamic semantic patterns in large-scale text, enriches the research paradigm of topic modeling, and supports complex text analysis tasks in multiple domains.

cs.CL↗

arXiv2025

Modular Task Decomposition and Dynamic Collaboration in Multi-Agent Systems Driven by Large Language Models

This paper addresses the limitations of a single agent in task decomposition and collaboration during complex task execution, and proposes a multi-agent architecture for modular task decomposition and dynamic collaboration based on large language models. The method first converts natural language task descriptions into unified semantic representations through a large language model. On this basis, a modular decomposition mechanism is introduced to break down the overall goal into multiple hierarchical sub-tasks. Then, dynamic scheduling and routing mechanisms enable reasonable division of labor and realtime collaboration among agents, allowing the system to adjust strategies continuously according to environmental feedback, thus maintaining efficiency and stability in complex tasks. Furthermore, a constraint parsing and global consistency mechanism is designed to ensure coherent connections between sub-tasks and balanced workload, preventing performance degradation caused by redundant communication or uneven resource allocation. The experiments validate the architecture across multiple dimensions, including task success rate, decomposition efficiency, sub-task coverage, and collaboration balance. The results show that the proposed method outperforms existing approaches in both overall performance and robustness, achieving a better balance between task complexity and communication overhead. In conclusion, this study demonstrates the effectiveness and feasibility of language-driven task decomposition and dynamic collaboration in multi-agent systems, providing a systematic solution for task execution in complex environments.

cs.AI↗

arXiv2025

Planets Across Space and Time (PAST). VIII : Kinematic Characterization and Identification of Radial Velocity Variables for the LAMOST-Gaia-TESS Stars

The Transiting Exoplanet Survey Satellite (TESS) has discovered over 6700 nearby exoplanets candidates using the transit method through its all-sky survey. Characterizing the kinematic properties and identifying variable stars for the TESS stellar sample is crucial for revealing the correlations between the properties of planetary systems and the properties of stars (e.g., Galactic components, age, chemistry, dynamics, radiation). Based on data from TESS, Gaia DR3, and LAMOST DR10, we present a catalog of kinematic properties (i.e., Galactic positions, velocities, orbits, Galactic components, and kinematic age) as well as other basic stellar parameters for $\sim 660,000$ TESS stars. Our analysis of the kinematic catalog reveals that stars belonging to different Galactic components (i.e., thin disk, thick disk, halo and 12 streams in the disk) display distinctive kinematic and chemical properties. We also find that hot planets with period less then 10 days in the TESS sample favor thin disk stars compared to thick disk stars, consistent with previous studies. Furthermore, using the LAMOST multiple-epoch observations, we identify 41,445 stars exhibiting significant radial velocity variations, among which 7,846 are classified as binary stars. By fitting the radial velocity curves, we further derive orbital parameters (e.g., mass ratio, orbital period and eccentricity) for 297 binaries. The observed decreasing orbital eccentricity with shorting period reveals evidence of tidal circularization. The catalogs constructed in this work have laid a solid foundation for future work on the formation and evolution of stellar and planetary systems in different Galactic environments.

astro-ph.EP↗

arXiv2025

Discontinuous Behavior of Time-of-Flight Distribution for Bi-impulsive Earth-Moon Transfers in the Three-Body Model

As interest in the Earth-Moon transfers renewed around the world, understanding the solution space of transfer trajectories facilitates the construction of transfers. This paper is devoted to reporting a novel or less-reported phenomenon about the solution space of bi-impulsive Earth-Moon transfers in the Earth-Moon planar circular restricted three-body problem. Differing from the previous works focusing on the transfer characteristics of the solution space, we focus on the distribution of the construction parameters, i.e., departure phase angle at the Earth parking orbit, initial-to-circular velocity ratio, and time of flight. Firstly, the construction method of bi-impulsive transfers is described, and the solutions satisfying the given constraints are obtained from the grid search method and trajectory correction. Then, the distribution of the obtained solutions is analyzed, and an interesting phenomenon about the discontinuous behavior of the time-of-flight distribution for each departure phase angle is observed and briefly reported. This phenomenon can further provide useful insight into the construction of bi-impulsive transfers, deepening the understanding of the corresponding solution space.

math.OC↗

arXiv2025

Greedy Sampling Is Provably Efficient for RLHF

Reinforcement Learning from Human Feedback (RLHF) has emerged as a key technique for post-training large language models. Despite its empirical success, the theoretical understanding of RLHF is still limited, as learning the KL-regularized target with only preference feedback poses additional challenges compared with canonical RL. Existing works mostly study the reward-based Bradley-Terry (BT) preference model, and extend classical designs utilizing optimism or pessimism. This work, instead, considers the general preference model (whose practical relevance has been observed recently) and obtains performance guarantees with major, order-wise improvements over existing ones. Surprisingly, these results are derived from algorithms that directly use the empirical estimates (i.e., greedy sampling), as opposed to constructing optimistic or pessimistic estimates in previous works. This insight has a deep root in the unique structural property of the optimal policy class under the KL-regularized target, and we further specialize it to the BT model, highlighting the surprising sufficiency of greedy sampling in RLHF.

cs.LG↗

arXiv2025

Structural Generalization for Microservice Routing Using Graph Neural Networks

This paper focuses on intelligent routing in microservice systems and proposes an end-to-end optimization framework based on graph neural networks. The goal is to improve routing decision efficiency and overall system performance under complex topologies. The method models invocation relationships among microservices as a graph. In this graph, service nodes and communication links are treated as graph nodes and edges. Multi-dimensional features such as node states, link latency, and call frequency are used as input. A multi-layer graph neural network is employed to perform high-order information aggregation and structural modeling. The model outputs a score for each candidate service path. These scores are then used to guide dynamic routing decisions. To improve the model's ability to assess path quality, an edge-aware attention mechanism is introduced. This mechanism helps the model capture instability and bottleneck risks in service communications more accurately. The paper also conducts a systematic analysis of the model's performance under different network depths, topology densities, and service scales. It evaluates the effectiveness of the method in terms of routing accuracy, prediction error, and system stability. Experimental results show that the proposed method outperforms existing mainstream strategies across multiple key metrics. It handles highly dynamic and concurrent microservice environments effectively and demonstrates strong performance, robustness, and structural generalization.

cs.NI↗

arXiv2025

Spatiotemporal Traffic Prediction in Distributed Backend Systems via Graph Neural Networks

This paper addresses the problem of traffic prediction in distributed backend systems and proposes a graph neural network based modeling approach to overcome the limitations of traditional models in capturing complex dependencies and dynamic features. The system is abstracted as a graph with nodes and edges, where node features represent traffic and resource states, and adjacency relations describe service interactions. A graph convolution mechanism enables multi order propagation and aggregation of node features, while a gated recurrent structure models historical sequences dynamically, thus integrating spatial structures with temporal evolution. A spatiotemporal joint modeling module further fuses graph representation with temporal dependency, and a decoder generates future traffic predictions. The model is trained with mean squared error to minimize deviations from actual values. Experiments based on public distributed system logs construct combined inputs of node features, topology, and sequences, and compare the proposed method with mainstream baselines using MSE, RMSE, MAE, and MAPE. Results show that the proposed method achieves stable performance and low error across different prediction horizons and model depths, significantly improving the accuracy and robustness of traffic forecasting in distributed backend systems and verifying the potential of graph neural networks in complex system modeling.

cs.DC↗

arXiv2025

Recursive Inference for Heterogeneous Multi-Output GP State-Space Models with Arbitrary Moment Matching

Accurate learning of system dynamics is becoming increasingly crucial for advanced control and decision-making in engineering. However, real-world systems often exhibit multiple channels and highly nonlinear transition dynamics, challenging traditional modeling methods. To enable online learning for these systems, this paper formulates the system as Gaussian process state-space models (GPSSMs) and develops a recursive learning method. The main contributions are threefold. First, a heterogeneous multi-output kernel is designed, allowing each output dimension to adopt distinct kernel types, hyperparameters, and input variables, improving expressiveness in multi-dimensional dynamics learning. Second, an inducing-point management algorithm enhances computational efficiency through independent selection and pruning for each output dimension. Third, a unified recursive inference framework for GPSSMs is derived, supporting general moment matching approaches, including the extended Kalman filter (EKF), unscented Kalman filter (UKF), and assumed density filtering (ADF), enabling accurate learning under strong nonlinearity and significant noise. Experiments on synthetic and real-world datasets show that the proposed method matches the accuracy of SOTA offline GPSSMs with only 1/100 of the runtime, and surpasses SOTA online GPSSMs by around 70% in accuracy under heavy noise while using only 1/20 of the runtime.

stat.ML↗

arXiv2025

LLMs as Scalable, General-Purpose Simulators For Evolving Digital Agent Training

Digital agents require diverse, large-scale UI trajectories to generalize across real-world tasks, yet collecting such data is prohibitively expensive in both human annotation, infra and engineering perspectives. To this end, we introduce $\textbf{UI-Simulator}$, a scalable paradigm that generates structured UI states and transitions to synthesize training trajectories at scale. Our paradigm integrates a digital world simulator for diverse UI states, a guided rollout process for coherent exploration, and a trajectory wrapper that produces high-quality and diverse trajectories for agent training. We further propose $\textbf{UI-Simulator-Grow}$, a targeted scaling strategy that enables more rapid and data-efficient scaling by prioritizing high-impact tasks and synthesizes informative trajectory variants. Experiments on WebArena and AndroidWorld show that UI-Simulator rivals or surpasses open-source agents trained on real UIs with significantly better robustness, despite using weaker teacher models. Moreover, UI-Simulator-Grow matches the performance of Llama-3-70B-Instruct using only Llama-3-8B-Instruct as the base model, highlighting the potential of targeted synthesis scaling paradigm to continuously and efficiently enhance the digital agents.

cs.CL↗

arXiv2025

Revisiting Trust in the Era of Generative AI: Factorial Structure and Latent Profiles

Trust is one of the most important factors shaping whether and how people adopt and rely on artificial intelligence (AI). Yet most existing studies measure trust in terms of functionality, focusing on whether a system is reliable, accurate, or easy to use, while giving less attention to the social and emotional dimensions that are increasingly relevant for today's generative AI (GenAI) systems. These systems do not just process information; they converse, respond, and collaborate with users, blurring the line between tool and partner. In this study, we introduce and validate the Human-AI Trust Scale (HAITS), a new measure designed to capture both the rational and relational aspects of trust in GenAI. Drawing on prior trust theories, qualitative interviews, and two waves of large-scale surveys in China and the United States, we used exploratory (n = 1,546) and confirmatory (n = 1,426) factor analyses to identify four key dimensions of trust: Affective Trust, Competence Trust, Benevolence & Integrity, and Perceived Risk. We then applied latent profile analysis to classify users into six distinct trust profiles, revealing meaningful differences in how affective-competence trust and trust-distrust frameworks coexist across individuals and cultures. Our findings offer a validated, culturally sensitive tool for measuring trust in GenAI and provide new insight into how trust evolves in human-AI interaction. By integrating instrumental and relational perspectives of trust, this work lays the foundation for more nuanced research and design of trustworthy AI systems.

cs.HC↗