arXiv subjects

Fei Du

Publications and source records attributed to Fei Du.

At least 19 recordsLinked to original sources

arXiv2026

SCAIR: Schema-Conditioned Agentic Iterative Reasoning for Enterprise Knowledge Graphs

Knowledge Graph-based Retrieval-Augmented Generation (KG-RAG) enables natural language interaction with structured enterprise knowledge, yet existing agentic approaches that perform well on public benchmarks often fail to generalize to real-world enterprise Knowledge Graphs (KGs), which are dense, schema-driven, and operationally constrained. To address these limitations, we propose SCAIR (Schema-Conditioned Agentic Iterative Reasoning), a training-free framework that integrates structured planning with controlled iterative reasoning by injecting schema-conditioned structural priors and enforcing schema-aware traversal during multi-hop reasoning. Experiments on an enterprise-oriented benchmark constructed from a real-world Configuration Management DataBase (CMDB) demonstrate that SCAIR substantially improves performance over existing KG-RAG methods. Crucially, our study highlights that reliable enterprise graph reasoning cannot rely on generic agentic designs; instead, it must explicitly incorporate the target domain's structural and operational constraints into the reasoning process. We demonstrate that by aligning agent design with business logic, substantial performance gains can be achieved without the need for costly model retraining.

cs.AI↗

arXiv2026

Lumos-Nexus: Efficient Frequency Bridging with Homogeneous Latent Space for Video Unified Models

Connector-based video unified models have demonstrated strong capability in instruction-grounded video synthesis, but integrating a large high-fidelity generator into the unified training loop is computationally prohibitive, limiting achievable visual quality. We therefore propose Lumos-Nexus, a training-efficient unified video generation framework that facilitates the development of strong reasoning-driven generation capabilities while significantly enhancing visual fidelity. Lumos-Nexus adopts a two-stage design: 1) During training, only a lightweight generator is aligned with the understanding block to learn to take in reasoning-driven semantic control. 2) During inference, we introduce Unified Progressive Frequency Bridging (UPFB) to progressively hand off generation to a high-capacity pretrained generator in the shared latent space, enabling coarse-to-fine refinement and producing high-fidelity videos without compromising reasoning quality. To fill the gap in reasoning-driven video generation benchmarks, we introduce VR-Bench, which assesses a model's capability to translate inferred intent into coherent and semantically aligned video content. Extensive experiments demonstrate that Lumos-Nexus achieves substantial gains in visual realism and temporal coherence on VBench, while exhibiting strong reasoning-based generative performance on VR-Bench. Code and models are available at https://jiazheng-xing.github.io/nexus-lumos-home/.

cs.CV↗

arXiv2026

LumosX: Relate Any Identities with Their Attributes for Personalized Video Generation

Recent advances in diffusion models have significantly improved text-to-video generation, enabling personalized content creation with fine-grained control over both foreground and background elements. However, precise face-attribute alignment across subjects remains challenging, as existing methods lack explicit mechanisms to ensure intra-group consistency. Addressing this gap requires both explicit modeling strategies and face-attribute-aware data resources. We therefore propose LumosX, a framework that advances both data and model design. On the data side, a tailored collection pipeline orchestrates captions and visual cues from independent videos, while multimodal large language models (MLLMs) infer and assign subject-specific dependencies. These extracted relational priors impose a finer-grained structure that amplifies the expressive control of personalized video generation and enables the construction of a comprehensive benchmark. On the modeling side, Relational Self-Attention and Relational Cross-Attention intertwine position-aware embeddings with refined attention dynamics to inscribe explicit subject-attribute dependencies, enforcing disciplined intra-group cohesion and amplifying the separation between distinct subject clusters. Comprehensive evaluations on our benchmark demonstrate that LumosX achieves state-of-the-art performance in fine-grained, identity-consistent, and semantically aligned personalized multi-subject video generation. Code and models are available at https://jiazheng-xing.github.io/lumosx-home/.

cs.CV↗

arXiv2025

Numerical study on wave attenuation via 2D fully kinetic electromagnetic particle-in-cell simulations

The propagation and absorption of electromagnetic waves in plasma is one of the fundamental issues in plasma physics. The electromagnetic particle-in-cell method with the finite-difference time-domain solver plus Monte Carlo collision model would be the most accurate method to simulate the wave-plasma interaction. However, the numerical effects of this method have not been carefully investigated especially in two dimensions. In this paper, the 2D PIC method is used to study the electromagnetic wave attenuation by fluorescent lamp plasma tubes. The study finds that the number of macro-particles and the incident electromagnetic wave amplitude have minor effects on the wave attenuation within a certain appropriate parameter range. Furthermore, the effects of electromagnetic wave frequency, the plasma distribution structures, and collision types on wave attenuation are investigated. Particularly, it is found that the staggered way of arranging the plasma distribution structures can achieve better wave attenuation than the parallel way, which agrees with our recent experimental observation.

physics.plasm-ph↗

arXiv2024

RealisID: Scale-Robust and Fine-Controllable Identity Customization via Local and Global Complementation

Recently, the success of text-to-image synthesis has greatly advanced the development of identity customization techniques, whose main goal is to produce realistic identity-specific photographs based on text prompts and reference face images. However, it is difficult for existing identity customization methods to simultaneously meet the various requirements of different real-world applications, including the identity fidelity of small face, the control of face location, pose and expression, as well as the customization of multiple persons. To this end, we propose a scale-robust and fine-controllable method, namely RealisID, which learns different control capabilities through the cooperation between a pair of local and global branches. Specifically, by using cropping and up-sampling operations to filter out face-irrelevant information, the local branch concentrates the fine control of facial details and the scale-robust identity fidelity within the face region. Meanwhile, the global branch manages the overall harmony of the entire image. It also controls the face location by taking the location guidance as input. As a result, RealisID can benefit from the complementarity of these two branches. Finally, by implementing our branches with two different variants of ControlNet, our method can be easily extended to handle multi-person customization, even only trained on single-person datasets. Extensive experiments and ablation studies indicate the effectiveness of RealisID and verify its ability in fulfilling all the requirements mentioned above.

cs.CV↗

arXiv2024

SHMT: Self-supervised Hierarchical Makeup Transfer via Latent Diffusion Models

This paper studies the challenging task of makeup transfer, which aims to apply diverse makeup styles precisely and naturally to a given facial image. Due to the absence of paired data, current methods typically synthesize sub-optimal pseudo ground truths to guide the model training, resulting in low makeup fidelity. Additionally, different makeup styles generally have varying effects on the person face, but existing methods struggle to deal with this diversity. To address these issues, we propose a novel Self-supervised Hierarchical Makeup Transfer (SHMT) method via latent diffusion models. Following a "decoupling-and-reconstruction" paradigm, SHMT works in a self-supervised manner, freeing itself from the misguidance of imprecise pseudo-paired data. Furthermore, to accommodate a variety of makeup styles, hierarchical texture details are decomposed via a Laplacian pyramid and selectively introduced to the content representation. Finally, we design a novel Iterative Dual Alignment (IDA) module that dynamically adjusts the injection condition of the diffusion model, allowing the alignment errors caused by the domain gap between content and makeup representations to be corrected. Extensive quantitative and qualitative analyses demonstrate the effectiveness of our method. Our code is available at \url{https://github.com/Snowfallingplum/SHMT}.

cs.CV↗

arXiv2024

LLM-Powered Explanations: Unraveling Recommendations Through Subgraph Reasoning

Recommender systems are pivotal in enhancing user experiences across various web applications by analyzing the complicated relationships between users and items. Knowledge graphs(KGs) have been widely used to enhance the performance of recommender systems. However, KGs are known to be noisy and incomplete, which are hard to provide reliable explanations for recommendation results. An explainable recommender system is crucial for the product development and subsequent decision-making. To address these challenges, we introduce a novel recommender that synergies Large Language Models (LLMs) and KGs to enhance the recommendation and provide interpretable results. Specifically, we first harness the power of LLMs to augment KG reconstruction. LLMs comprehend and decompose user reviews into new triples that are added into KG. In this way, we can enrich KGs with explainable paths that express user preferences. To enhance the recommendation on augmented KGs, we introduce a novel subgraph reasoning module that effectively measures the importance of nodes and discovers reasoning for recommendation. Finally, these reasoning paths are fed into the LLMs to generate interpretable explanations of the recommendation results. Our approach significantly enhances both the effectiveness and interpretability of recommender systems, especially in cross-selling scenarios where traditional methods falter. The effectiveness of our approach has been rigorously tested on four open real-world datasets, with our methods demonstrating a superior performance over contemporary state-of-the-art techniques by an average improvement of 12%. The application of our model in a multinational engineering and technology company cross-selling recommendation system further underscores its practical utility and potential to redefine recommendation practices through improved accuracy and user trust.

cs.IR↗

arXiv2024

Observability of gravitational waves excited by binary stars orbiting around a supermassive black hole by space-based gravitational wave observatory

We produce the gravitational waveforms for the extreme mass ratio inspiral systems (EMRIs) of binary stars moving around central supermassive black hole (SBH), or called B-EMRIs. We calculate the external orbits of the binary stars via the commonly used Hamilton-Jacobi (HJ) approach, and calculate the internal orbits of the binary stars via Lagrangian approach. To improve accuracy we adopt the quadrupole-octupole expression of gravitational wave (GW) and study the contribution of radiation reaction. Compared to the waveforms of EMRIs, there are higher frequency oscillations superposed on the waveforms of B-EMRIs. We perform frequency spectrum analysis of the GW waveforms, and find that higher frequency signals give their prominency in the waveforms of B-EMRIs. To obtain high precise result for future observation of GWs from space-based detector, we take into account gravito-electromagnetic (GEM) force, and compare the waveforms of B-EMRIs with GEM effects against those of B-EMRIs without GEM effects and against those of EMRIs. The result of mismatch shows that the waveforms of B-EMRIs are credibly distinguishable by the space-based GW detectors when GEM force is considered.

gr-qc↗

arXiv2024

Free-standing cubic gauche nitrogen stable at 760 K under ambient pressure

Cubic gauche nitrogen (cg-N) has received wide attention due to its high energy density and environmental friendliness. However, existing synthesis methods for cg-N predominantly rely on the high-pressure techniques, or the utilization of nanoconfined effects using highly toxic and sensitive sodium azide as precursor, which significantly restrict the practical application of cg-N as high energy density materials (HDEM). Here, based on the first-principles simulations, we find that the adsorption of potassium on the cg-N surface exhibits superior stabilization compared to sodium. Then, we chose the safer potassium azide as raw material for synthesizing cg-N. Through plasma-enhanced chemical vapor deposition treatment, the free-standing cg-N was successfully synthesized without the need of high-pressure and nanoconfined effects. Importantly, it demonstrates excellent thermal stability up to 760 K, and then a rapid and intense thermal decomposition occurs, exhibiting typical behaviors of HDEM thermal decomposition. Our work has significantly promoted the practical application of cg-N as HDEM.

cond-mat.mtrl-sci↗

arXiv2023

Data Pruning via Moving-one-Sample-out

In this paper, we propose a novel data-pruning approach called moving-one-sample-out (MoSo), which aims to identify and remove the least informative samples from the training set. The core insight behind MoSo is to determine the importance of each sample by assessing its impact on the optimal empirical risk. This is achieved by measuring the extent to which the empirical risk changes when a particular sample is excluded from the training set. Instead of using the computationally expensive leaving-one-out-retraining procedure, we propose an efficient first-order approximator that only requires gradient information from different training stages. The key idea behind our approximation is that samples with gradients that are consistently aligned with the average gradient of the training set are more informative and should receive higher scores, which could be intuitively understood as follows: if the gradient from a specific sample is consistent with the average gradient vector, it implies that optimizing the network using the sample will yield a similar effect on all remaining samples. Experimental results demonstrate that MoSo effectively mitigates severe performance degradation at high pruning ratios and achieves satisfactory performance across various settings.

cs.LG↗

arXiv2023

SwinRDM: Integrate SwinRNN with Diffusion Model towards High-Resolution and High-Quality Weather Forecasting

Data-driven medium-range weather forecasting has attracted much attention in recent years. However, the forecasting accuracy at high resolution is unsatisfactory currently. Pursuing high-resolution and high-quality weather forecasting, we develop a data-driven model SwinRDM which integrates an improved version of SwinRNN with a diffusion model. SwinRDM performs predictions at 0.25-degree resolution and achieves superior forecasting accuracy to IFS (Integrated Forecast System), the state-of-the-art operational NWP model, on representative atmospheric variables including 500 hPa geopotential (Z500), 850 hPa temperature (T850), 2-m temperature (T2M), and total precipitation (TP), at lead times of up to 5 days. We propose to leverage a two-step strategy to achieve high-resolution predictions at 0.25-degree considering the trade-off between computation memory and forecasting accuracy. Recurrent predictions for future atmospheric fields are firstly performed at 1.40625-degree resolution, and then a diffusion-based super-resolution model is leveraged to recover the high spatial resolution and finer-scale atmospheric details. SwinRDM pushes forward the performance and potential of data-driven models for a large margin towards operational applications.

cs.AI↗

arXiv2023

Global and Local Mixture Consistency Cumulative Learning for Long-tailed Visual Recognitions

In this paper, our goal is to design a simple learning paradigm for long-tail visual recognition, which not only improves the robustness of the feature extractor but also alleviates the bias of the classifier towards head classes while reducing the training skills and overhead. We propose an efficient one-stage training strategy for long-tailed visual recognition called Global and Local Mixture Consistency cumulative learning (GLMC). Our core ideas are twofold: (1) a global and local mixture consistency loss improves the robustness of the feature extractor. Specifically, we generate two augmented batches by the global MixUp and local CutMix from the same batch data, respectively, and then use cosine similarity to minimize the difference. (2) A cumulative head tail soft label reweighted loss mitigates the head class bias problem. We use empirical class frequencies to reweight the mixed label of the head-tail class for long-tailed data and then balance the conventional loss and the rebalanced loss with a coefficient accumulated by epochs. Our approach achieves state-of-the-art accuracy on CIFAR10-LT, CIFAR100-LT, and ImageNet-LT datasets. Additional experiments on balanced ImageNet and CIFAR demonstrate that GLMC can significantly improve the generalization of backbones. Code is made publicly available at https://github.com/ynu-yangpeng/GLMC.

cs.CV↗

arXiv2022

FAKD: Feature Augmented Knowledge Distillation for Semantic Segmentation

In this work, we explore data augmentations for knowledge distillation on semantic segmentation. To avoid over-fitting to the noise in the teacher network, a large number of training examples is essential for knowledge distillation. Imagelevel argumentation techniques like flipping, translation or rotation are widely used in previous knowledge distillation framework. Inspired by the recent progress on semantic directions on feature-space, we propose to include augmentations in feature space for efficient distillation. Specifically, given a semantic direction, an infinite number of augmentations can be obtained for the student in the feature space. Furthermore, the analysis shows that those augmentations can be optimized simultaneously by minimizing an upper bound for the losses defined by augmentations. Based on the observation, a new algorithm is developed for knowledge distillation in semantic segmentation. Extensive experiments on four semantic segmentation benchmarks demonstrate that the proposed method can boost the performance of current knowledge distillation methods without any significant overhead. Code is available at: https://github.com/jianlong-yuan/FAKD.

cs.CV↗

arXiv2021

1st Place Solution to ECCV-TAO-2020: Detect and Represent Any Object for Tracking

We extend the classical tracking-by-detection paradigm to this tracking-any-object task. Solid detection results are first extracted from TAO dataset. Some state-of-the-art techniques like \textbf{BA}lanced-\textbf{G}roup \textbf{S}oftmax (\textbf{BAGS}\cite{li2020overcoming}) and DetectoRS\cite{qiao2020detectors} are integrated during detection. Then we learned appearance features to represent any object by training feature learning networks. We ensemble several models for improving detection and feature representation. Simple linking strategies with most similar appearance features and tracklet-level post association module are finally applied to generate final tracking results. Our method is submitted as \textbf{AOA} on the challenge website. Code is available at https://github.com/feiaxyt/Winner_ECCV20_TAO.

cs.CV↗

arXiv2015

Pressure-induced Superconductivity in the Iron-based Ladder Material BaFe2S3

All the iron-based superconductors identified to date share a square lattice composed of Fe atoms as a common feature, despite having different crystal structures. In copper-based materials, the superconducting phase emerges not only in square lattice structures but also in ladder structures. Yet iron-based superconductors without a square lattice motif have not been found despite being actively sought out. Here, we report the discovery of pressure-induced superconductivity in the iron-based spin-ladder material BaFe2S3, a Mott insulator with striped-type magnetic ordering below ~120 K. On the application of pressure this compound exhibits a metal-insulator transition at about 11 GPa, followed by the appearance of superconductivity below Tc = 14 K, right after the onset of the metallic phase. Our findings indicate that iron-based ladder compounds represent promising material platforms, in particular for studying the fundamentals of iron-based superconductivity.

cond-mat.supr-con↗

arXiv2015

Electronic structure of Li$_{1+x}$[Mn$_{0.5}$Ni$_{0.5}$]$_{1-x}$O$_2$ studied by photoemission and x-ray absorption spectroscopy

We have studied the electronic structure of Li$_{1+x}$[Mn$_{0.5}$Ni$_{0.5}$]$_{1-x}$O$_2$ ($x$ = 0.00 and 0.05), one of the promising cathode materials for Li ion battery, by means of x-ray photoemission and absorption spectroscopy. The results show that the valences of Mn and Ni are basically 4+ and 2+, respectively. However, the Mn$^{3+}$ component in the $x$ = 0.00 sample gradually increases with the bulk sensitivity of the experiment, indicating that the Jahn-Teller active Mn$^{3+}$ ions are introduced in the bulk due to the site exchange between Li and Ni. The Mn$^{3+}$ component gets negligibly small in the $x$ = 0.05 sample, which indicates that the excess Li suppresses the site exchange and removes the Jahn-Teller active Mn$^{3+}$.

cond-mat.str-el↗

arXiv2015

Temperature and composition phase diagram in the iron-based ladder compounds Ba1-xCsxFe2Se3

We investigated the iron-based ladder compounds (Ba,Cs)Fe$_2$Se$_3$. Their parent compounds, BaFe$_2$Se$_3$ and CsFe$_2$Se$_3$, have different space groups, formal valences of Fe and magnetic structures. Electrical resistivity, specific heat, magnetic susceptibility, X-ray diffraction and powder neutron diffraction measurements were conducted to obtain temperature and composition phase diagram of this system. Block magnetism observed in BaFe$_2$Se$_3$ is drastically suppressed with Cs doping. In contrast, stripe magnetism observed in CsFe$_2$Se$_3$ is not so fragile against Ba doping. New type of magnetic structure appears in intermediate compositions, which is similar to stripe magnetism of CsFe$_2$Se$_3$, but inter-ladder spin configuration is different. Intermediate compounds show insulating behavior, nevertheless finite $T$-linear contribution in specific heat was obtained at low temperatures.

cond-mat.str-el↗

arXiv2012

Block magnetism coupled with local distortion in the iron-based spin-ladder compound BaFe2Se3

Magnetism in the insulating BaFe$_2$Se$_3$ was examined through susceptibility, specific heat, resistivity and neutron diffraction measurements. After formation of a short-range magnetic correlation, a long-range ordering was observed below $T_{\rm N}\sim 255$ K. The transition is obscured by bulk properties. Magnetic moments ($\parallel a$) are arranged to form a Fe$_4$ ferromagnetic unit, and each Fe$_4$ stacks antiferromagnetically. This block magnetism is of the third type among magnetic structures of ferrous materials. The magnetic ordering drives unusually large distortion via magnetoelastic coupling.

cond-mat.str-el↗