arXiv subjects

Chen Cui

Publications and source records attributed to Chen Cui.

At least 19 recordsLinked to original sources

arXiv2026

Nemotron 3 Ultra: Open, Efficient Mixture-of-Experts Hybrid Mamba-Transformer Model for Agentic Reasoning

We introduce Nemotron 3 Ultra, a 550 billion total and 55 billion active parameter Mixture-of-Experts Hybrid Mamba-Attention language model. We pre-trained Nemotron 3 Ultra on 20 trillion text tokens, then extended the context length to 1M tokens, and post-trained using Supervised Fine Tuning (SFT), Reinforcement Learning (RL), and Multi-teacher On-Policy Distillation (MOPD). Nemotron 3 Ultra is our most capable model yet, employing multiple key technologies - LatentMoE, Multi Token Prediction (MTP), NVFP4 pre-training, multi-environment RLVR, MOPD, and reasoning budget control. Nemotron 3 Ultra achieves up to ~6x higher inference throughput as compared to state-of-the-art publicly available LLMs while attaining on-par accuracy. The state-of-the-art accuracy, high inference throughput, and 1M token context length make Nemotron 3 Ultra ideal for long-running autonomous agentic tasks. We open-source the base, post-trained, and quantized checkpoints, along with the training data and recipe on HuggingFace.

cs.CL↗

arXiv2026

Nemotron 3 Nano Omni: Efficient and Open Multimodal Intelligence

We introduce Nemotron 3 Nano Omni, the latest model in the Nemotron multimodal series and the first to natively support audio inputs alongside text, images, and video. Nemotron 3 Nano Omni delivers consistent accuracy improvements over its predecessor, Nemotron Nano V2 VL, across all modalities, enabled by advances in architecture, training data and recipes. In particular, Nemotron 3 delivers leading results in real-world document understanding, long audio-video comprehension, and agentic computer use. Built on the highly efficient Nemotron 3 Nano 30B-A3B backbone, Nemotron 3 Nano Omni further incorporates innovative multimodal token-reduction techniques to deliver substantially lower inference latency and higher throughput than other models of similar size. We are releasing model checkpoints in BF16, FP8, and FP4 formats, along with portions of the training data and codebase to facilitate further research and development.

cs.LG↗

arXiv2025

Projected Sobolev Natural Gradient Descent for Efficient Neural Network Solution of the Gross--Pitaevskii Equation

This paper introduces a projected Sobolev natural gradient descent (NGD) method for computing ground states of the Gross--Pitaevskii equation. By projecting a continuous Riemannian Sobolev gradient flow onto the normalized neural network tangent space, we derive a discrete NGD algorithm that preserves the normalization constraint. The numerical implementation employs variational Monte Carlo with a hybrid sampling strategy to accurately account for the normalization constant. To enhance computational efficiency, a matrix-free Nystr\"om-preconditioned conjugate gradient solver is adopted to approximate the NGD operator without explicit matrix assembly. Numerical experiments demonstrate that the proposed method converges significantly faster than physics-informed neural network approaches and exhibits near-constant wall-clock time and memory over the tested high-dimensional range when the hidden architecture is fixed and the implicit Gram operator is used. Moreover, the resulting neural-network solutions provide high-quality initial guesses that substantially accelerate subsequent refinement by traditional high-precision solvers.

math.NA↗

arXiv2025

Nodal Hybrid Neural Solvers for Parametric PDE Systems

The numerical solution of partial differential equations (PDEs) is fundamental to scientific and engineering computing. In the presence of strong anisotropy, material heterogeneity, and complex geometries, however, classical iterative solvers often suffer from reduced efficiency and require substantial problem-dependent tuning. The Fourier neural solver (FNS) is a learning-based hybrid iterative solver for such problems without extensive manual parameter tuning, but its original design is primarily effective for scalar PDEs on structured meshes and is difficult to extend directly to unstructured meshes or strongly coupled PDE systems. Building on the FNS framework, we introduce block smoothing operators and graph neural networks to construct a solver for unstructured systems, termed the graph Fourier neural solver (G-FNS). We further incorporate a coordinate transformation network to develop the adaptive graph Fourier neural solver (AG-FNS), and then extend this formulation to a frequency-domain multilevel variant, ML-AG-FNS. Rigorous analysis shows that, under suitable mathematical assumptions, the proposed method achieves mesh-independent convergence rate. Error-spectrum visualizations further indicate that AG-FNS can capture complex multiscale error modes. Extensive experiments on two-dimensional anisotropic diffusion and on two- and three-dimensional isotropic/anisotropic linear elasticity problems over unstructured meshes demonstrate strong robustness and efficiency. The proposed framework can be used either as a solver or as a preconditioner for Krylov subspace methods. Overall, it substantially extends the original FNS methodology and broadens the applicability of this class of neural solvers.

math.NA↗

arXiv2025

A full breakthrough in vacuum ultraviolet nonlinear optical performance of NH4B4O6F

The lack of suitable vacuum ultraviolet (VUV) nonlinear optical (NLO) crystals has hindered the development of compact, high-power VUV sources via second harmonic generation (SHG). Here, we report on the development of the fluorooxoborate crystal NH4B4O6F (ABF) as a promising material for VUV light generation. For the first time, devices with specific phase-matching angles were constructed, achieving a record 158.9 nm VUV light through phase-matching SHG and a maximum nanosecond pulse energy of 4.8 mJ at 177.3 nm with a conversion efficiency of 5.9 %. The enhanced NLO performance is attributed to optimized arrangements of fluorine-based units creating asymmetric sublattices. This work marks a significant milestone in the field of NLO materials, facilitating the future applications of compact, high-power VUV lasers utilizing ABF.

cond-mat.mtrl-sci↗

arXiv2024

Momentum-Accelerated Richardson(m) and Their Multilevel Neural Solvers

Recently, designing neural solvers for large-scale linear systems of equations has emerged as a promising approach in scientific and engineering computing. This paper first introduce the Richardson(m) neural solver by employing a meta network to predict the weights of the long-step Richardson iterative method. Next, by incorporating momentum and preconditioning techniques, we further enhance convergence. Numerical experiments on anisotropic second-order elliptic equations demonstrate that these new solvers achieve faster convergence and lower computational complexity compared to both the Chebyshev iterative method with optimal weights and the Chebyshev semi-iteration method. To address the strong dependence of the aforementioned single-level neural solvers on PDE parameters and grid size, we integrate them with two multilevel neural solvers developed in recent years. Using alternating optimization techniques, we construct Richardson(m)-FNS for anisotropic equations and NAG-Richardson(m)-WANS for the Helmholtz equation. Numerical experiments show that these two multilevel neural solvers effectively overcome the drawback of single-level methods, providing better robustness and computational efficiency.

math.NA↗

arXiv2024

A Hybrid Iterative Neural Solver Based on Spectral Analysis for Parametric PDEs

Deep learning-based hybrid iterative methods (DL-HIM) have emerged as a promising approach for designing fast neural solvers to tackle large-scale sparse linear systems. DL-HIM combine the smoothing effect of simple iterative methods with the spectral bias of neural networks, which allows them to effectively eliminate both high-frequency and low-frequency error components. However, their efficiency may decrease if simple iterative methods can not provide effective smoothing, making it difficult for the neural network to learn mid-frequency and high-frequency components. This paper first conducts a convergence analysis for general DL-HIM from a spectral viewpoint, concluding that under reasonable assumptions, DL-HIM exhibit a convergence rate independent of grid size $h$ and physical parameters $\boldsymbol{\mu}$. To meet these assumptions, we design a neural network from an eigen perspective, focusing on learning the eigenvalues and eigenvectors corresponding to error components that simple iterative methods struggle to eliminate. Specifically, the eigenvalues are learned by a meta subnet, while the eigenvectors are approximated using Fourier modes with a transition matrix provided by another meta subnet. The resulting DL-HIM, termed the Fourier Neural Solver (FNS), can be trained to achieve a convergence rate independent of PDE parameters and grid size within a local neighborhood of the training scale by designing a loss function that ensures the neural network complements the smoothing effect of the damped Jacobi iterative methods. We verify the performance of FNS on five types of linear parametric PDEs.

math.NA↗

arXiv2024

A Neural Multigrid Solver for Helmholtz Equations with High Wavenumber and Heterogeneous Media

In this paper, we propose a deep learning-enhanced multigrid solver for high-frequency and heterogeneous Helmholtz equations. By applying spectral analysis, we categorize the iteration error into characteristic and non-characteristic components. We eliminate the non-characteristic components by a multigrid wave cycle, which employs carefully selected smoothers on each grid. We diminish the characteristic components by a learned phase function and the approximate solution of an advection-diffusion-reaction (ADR) equation, which is solved using another multigrid V-cycle on a coarser scale, referred to as the ADR cycle. The resulting solver, termed Wave-ADR-NS, enables the handling of error components with varying frequencies and overcomes constraints on the number of grid points per wavelength on coarse grids. Furthermore, we provide an efficient implementation using differentiable programming, making Wave-ADR-NS an end-to-end Helmholtz solver that incorporates parameters learned through a semi-supervised training. Wave-ADR-NS demonstrates robust generalization capabilities for both in-distribution and out-of-distribution velocity fields of varying difficulty. Comparative experiments with other multigrid methods validate its superior performance in solving heterogeneous 2D Helmholtz equations with wavenumbers exceeding 2000.

math.NA↗

arXiv2023

Implementing BERT and fine-tuned RobertA to detect AI generated news by ChatGPT

The abundance of information on social media has increased the necessity of accurate real-time rumour detection. Manual techniques of identifying and verifying fake news generated by AI tools are impracticable and time-consuming given the enormous volume of information generated every day. This has sparked an increase in interest in creating automated systems to find fake news on the Internet. The studies in this research demonstrate that the BERT and RobertA models with fine-tuning had the best success in detecting AI generated news. With a score of 98%, tweaked RobertA in particular showed excellent precision. In conclusion, this study has shown that neural networks can be used to identify bogus news AI generation news created by ChatGPT. The RobertA and BERT models' excellent performance indicates that these models can play a critical role in the fight against misinformation.

cs.CL↗

arXiv2022

Fourier Neural Solver for large sparse linear algebraic systems

Large sparse linear algebraic systems can be found in a variety of scientific and engineering fields, and many scientists strive to solve them in an efficient and robust manner. In this paper, we propose an interpretable neural solver, the Fourier Neural Solver (FNS), to address them. FNS is based on deep learning and Fast Fourier transform. Because the error between the iterative solution and the ground truth involves a wide range of frequency modes, FNS combines a stationary iterative method and frequency space correction to eliminate different components of the error. Local Fourier analysis reveals that the FNS can pick up on the error components in frequency space that are challenging to eliminate with stationary methods. Numerical experiments on the anisotropy diffusion equation, convection-diffusion equation, and Helmholtz equation show that FNS is more efficient and more robust than the state-of-the-art neural solver.

math.NA↗

arXiv2022

Testing the Ability of Language Models to Interpret Figurative Language

Figurative and metaphorical language are commonplace in discourse, and figurative expressions play an important role in communication and cognition. However, figurative language has been a relatively under-studied area in NLP, and it remains an open question to what extent modern language models can interpret nonliteral phrases. To address this question, we introduce Fig-QA, a Winograd-style nonliteral language understanding task consisting of correctly interpreting paired figurative phrases with divergent meanings. We evaluate the performance of several state-of-the-art language models on this task, and find that although language models achieve performance significantly over chance, they still fall short of human performance, particularly in zero- or few-shot settings. This suggests that further work is needed to improve the nonliteral reasoning capabilities of language models.

cs.CL↗

arXiv2021

Solving time-dependent parametric PDEs by multiclass classification-based reduced order model

In this paper, we propose a network model, the multiclass classification-based reduced order model (MC-ROM), for solving time-dependent parametric partial differential equations (PPDEs). This work is inspired by the observation of applying the deep learning-based reduced order model (DL-ROM) to solve diffusion-dominant PPDEs. We find that the DL-ROM has a good approximation for some special model parameters, but it cannot approximate the drastic changes of the solution as time evolves. Based on this fact, we classify the dataset according to the magnitude of the solutions and construct corresponding subnets dependent on different types of data. Then we train a classifier to integrate different subnets together to obtain the MC-ROM. When subsets have the same architecture, we can use transfer learning techniques to accelerate offline training. Numerical experiments show that the MC-ROM improves the generalization ability of the DL-ROM both for diffusion- and convection-dominant problems, and maintains the DL-ROM's advantage of good approximation ability. We also compare the approximation accuracy and computational efficiency of the proper orthogonal decomposition (POD) which is not suitable for convection-dominant problems. For diffusion-dominant problems, the MC-ROM has better approximation accuracy than the POD in a small dimensionality reduction space, and its computational performance is more efficient than the POD.

math.NA↗

arXiv2021

Graph-Fraudster: Adversarial Attacks on Graph Neural Network Based Vertical Federated Learning

Graph neural network (GNN) has achieved great success on graph representation learning. Challenged by large scale private data collected from user-side, GNN may not be able to reflect the excellent performance, without rich features and complete adjacent relationships. Addressing the problem, vertical federated learning (VFL) is proposed to implement local data protection through training a global model collaboratively. Consequently, for graph-structured data, it is a natural idea to construct a GNN based VFL framework, denoted as GVFL. However, GNN has been proved vulnerable to adversarial attacks. Whether the vulnerability will be brought into the GVFL has not been studied. This is the first study of adversarial attacks on GVFL. A novel adversarial attack method is proposed, named Graph-Fraudster. It generates adversarial perturbations based on the noise-added global node embeddings via the privacy leakage and the gradient of pairwise node. Specifically, first, Graph-Fraudster steals the global node embeddings and sets up a shadow model of the server for the attack generator. Second, noise is added into node embeddings to confuse the shadow model. At last, the gradient of pairwise node is used to generate attacks with the guidance of noise-added node embeddings. Extensive experiments on five benchmark datasets demonstrate that Graph-Fraudster achieves the state-of-the-art attack performance compared with baselines in different GNN based GVFLs. Furthermore, Graph-Fraudster can remain a threat to GVFL even if two possible defense mechanisms are applied. Additionally, some suggestions are put forward for the future work to improve the robustness of GVFL. The code and datasets can be downloaded at https://github.com/hgh0545/Graph-Fraudster.

cs.LG↗

arXiv2021

Salient Feature Extractor for Adversarial Defense on Deep Neural Networks

Recent years have witnessed unprecedented success achieved by deep learning models in the field of computer vision. However, their vulnerability towards carefully crafted adversarial examples has also attracted the increasing attention of researchers. Motivated by the observation that adversarial examples are due to the non-robust feature learned from the original dataset by models, we propose the concepts of salient feature(SF) and trivial feature(TF). The former represents the class-related feature, while the latter is usually adopted to mislead the model. We extract these two features with coupled generative adversarial network model and put forward a novel detection and defense method named salient feature extractor (SFE) to defend against adversarial attacks. Concretely, detection is realized by separating and comparing the difference between SF and TF of the input. At the same time, correct labels are obtained by re-identifying SF to reach the purpose of defense. Extensive experiments are carried out on MNIST, CIFAR-10, and ImageNet datasets where SFE shows state-of-the-art results in effectiveness and efficiency compared with baselines. Furthermore, we provide an interpretable understanding of the defense and detection process.

cs.CV↗

arXiv2021

GraphAttacker: A General Multi-Task GraphAttack Framework

Graph neural networks (GNNs) have been successfully exploited in graph analysis tasks in many real-world applications. The competition between attack and defense methods also enhances the robustness of GNNs. In this competition, the development of adversarial training methods put forward higher requirement for the diversity of attack examples. By contrast, most attack methods with specific attack strategies are difficult to satisfy such a requirement. To address this problem, we propose GraphAttacker, a novel generic graph attack framework that can flexibly adjust the structures and the attack strategies according to the graph analysis tasks. GraphAttacker generates adversarial examples through alternate training on three key components: the multi-strategy attack generator (MAG), the similarity discriminator (SD), and the attack discriminator (AD), based on the generative adversarial network (GAN). Furthermore, we introduce a novel similarity modification rate SMR to conduct a stealthier attack considering the change of node similarity distribution. Experiments on various benchmark datasets demonstrate that GraphAttacker can achieve state-of-the-art attack performance on graph analysis tasks of node classification, graph classification, and link prediction, no matter the adversarial training is conducted or not. Moreover, we also analyze the unique characteristics of each task and their specific response in the unified attack framework. The project code is available at https://github.com/honoluluuuu/GraphAttacker.

cs.LG↗

arXiv2020

MLANE: Meta-Learning Based Adaptive Network Embedding

Most existing random walk based network embedding methods often follow only one of two principles, homophily or structural equivalence. In real world networks, however, nodes exhibit a mixture of homophily and structural equivalence, which requires adaptive network embedding that can adaptively preserve both homophily and structural equivalence for different nodes in different down-stream analysis tasks. In this paper, we propose a novel method called Meta-Learning based Adaptive Network Embedding (MLANE), which can learn adaptive sampling strategy for different nodes in different tasks by incorporating sampling strategy learning with embedding learning into one optimization problem that can be solved via an end-to-end meta-learning framework. In extensive experiments on real datasets, MLANE shows significant performance improvements over the baselines. The source code of MLANE and the datasets used in experiments and all the hyperparameter settings for baselines are available at https://github.com/7733com/MLANE.

cs.SI↗

arXiv2020

SUPER: A Novel Lane Detection System

AI-based lane detection algorithms were actively studied over the last few years. Many have demonstrated superior performance compared with traditional feature-based methods. The accuracy, however, is still generally in the low 80% or high 90%, or even lower when challenging images are used. In this paper, we propose a real-time lane detection system, called Scene Understanding Physics-Enhanced Real-time (SUPER) algorithm. The proposed method consists of two main modules: 1) a hierarchical semantic segmentation network as the scene feature extractor and 2) a physics enhanced multi-lane parameter optimization module for lane inference. We train the proposed system using heterogeneous data from Cityscapes, Vistas and Apollo, and evaluate the performance on four completely separate datasets (that were never seen before), including Tusimple, Caltech, URBAN KITTI-ROAD, and X-3000. The proposed approach performs the same or better than lane detection models already trained on the same dataset and performs well even on datasets it was never trained on. Real-world vehicle tests were also conducted. Preliminary test results show promising real-time lane-detection performance compared with the Mobileye.

cs.CV↗

arXiv2019

Multi-View Fuzzy Clustering with The Alternative Learning between Shared Hidden Space and Partition

As the multi-view data grows in the real world, multi-view clus-tering has become a prominent technique in data mining, pattern recognition, and machine learning. How to exploit the relation-ship between different views effectively using the characteristic of multi-view data has become a crucial challenge. Aiming at this, a hidden space sharing multi-view fuzzy clustering (HSS-MVFC) method is proposed in the present study. This method is based on the classical fuzzy c-means clustering model, and obtains associ-ated information between different views by introducing shared hidden space. Especially, the shared hidden space and the fuzzy partition can be learned alternatively and contribute to each other. Meanwhile, the proposed method uses maximum entropy strategy to control the weights of different views while learning the shared hidden space. The experimental result shows that the proposed multi-view clustering method has better performance than many related clustering methods.

cs.LG↗