检索结果-内蒙古大学图书馆

Neural P3M: a long-range interaction modeling enhancer for geometric GNNs 24

学校读者我要写书评

暂无评论

Neural P3M: a long-range interaction modeling enhancer for g...

Proceedings of the 38th International Conference on Neural information Processing Systems

作者： Yusong Wang Chaoran Cheng Shaoning Li Yuxuan Ren Bin Shao Ge Liu Pheng-Ann Heng Nanning Zheng National Key Laboratory of Human-Machine Hybrid Augmented Intelligence National Engineering Research Center for Visual Information and Applications and Institute of Artificial Intelligence and Robotics Xi'an Jiaotong University University of Illinois Urbana-Champaign Department of Computer Science and Engineering The Chinese University of Hong Kong University of Science and Technology of China Microsoft Research AI4Science

ISBN: (纸本)9798331314385

Geometric graph neural networks (GNNs) have emerged as powerful tools for modeling molecular geometry. However, they encounter limitations in effectively capturing long-range interactions in large molecular systems due to the localization assumption of GNN. To address this challenge, we introduce Neural P3M, a versatile enhancer of geometric GNNs to expand the scope of their capabilities by incorporating mesh points alongside atoms and reimaging traditional mathematical operations in a trainable manner. Neural P3M exhibits flexibility across a wide range of molecular systems and demonstrates remarkable accuracy in predicting energies and forces, outperforming on benchmarks such as the MD22 dataset. It also achieves an average improvement of 22% on the OE62 dataset while integrating with various architectures. Codes are available at https://***/OnlyLoveKFC/Neural_P3M.

关键词：

Long-term and short-term plasticity independently mimicked in highly reliable Ru-doped Ge_(2)Sb_(2)Te_(5) electronic synapses

学校读者我要写书评

暂无评论

InfoMat 2024年第8期6卷 85-96页

作者： Qiang Wang Yachuan Wang Yankun Wang Luyue Jiang Jinyan Zhao Zhitang Song Jinshun Bi Libo Zhao Zhuangde Jiang Jutta Schwarzkopf Shengli Wu Bin Zhang Wei Ren Sannian Song Gang Niu State Key Laboratory for Manufacturing Systems Engineering Electronic Materials Research LaboratoryKey Laboratory of the Ministry of Education&International Center for Dielectric ResearchSchool of Electronic Science and EngineeringXi'an Jiaotong UniversityXi'anthe People's Republic of China School of Integrated Circuits Peking UniversityBeijingthe People's Republic of China Beijing Microelectronics Technology Institute Beijingthe People's Republic of China National Key Laboratory of Human-Machine Hybrid Augmented Intelligence National Engineering Research Center for Visual information and Applicationsand School of SoftwareXi'an Jiaotong UniversityXi'anthe People's Republic of China State Key Laboratory of Functional Materials for Informatics Shanghai Institute of Microsystem and Information TechnologyChinese Academy of SciencesShanghaithe People's Republic of China Key Laboratory of Microelectronics Device and Integrated Technology The Institute of Microelectronics of Chinese Academy of SciencesBeijingthe People's Republic of China University of Chinese Academy of Sciences Beijingthe People's Republic of China The State Key Laboratory for Manufacturing Systems Engineering&The International Joint Laboratory for Micro/Nano Manufacturing and Measurement Technology Xi'an Jiaotong UniversityXi'anthe People's Republic of China Leibniz-Institut für Kristallzüchtung Max-Born-Straße 2BerlinGermany Key Laboratory of Physical Electronics and Devices Ministry of EducationSchool of Electronic Science and EngineeringXi'an Jiaotong UniversityXi'anthe People's Republic of China The International Joint Laboratory for Micro/Nano Manufacturing and Measurement Technology Xi'an Jiaotong UniversityXi'anthe People's Republic of China

In order to fulfill the complex cognitive behaviors in neuromorphic systems with reduced peripheral circuits,the reliable electronic synapses mimicked by single device that achieves diverse long-term and short-term plasticity are *** change random access memory(PCRAM)is of great potential for artificial synapses,which faces,however,difficulty to realize short-term plasticity due to the long-lasting resistance *** work reports the ruthenium-doped Ge_(2)Sb_(2)Te_(5)(RuGST)based PCRAM,demonstrating a series of synaptic behaviors of short-term potentiation,pair-pulse facilitation,longterm depression,and short-term plasticity in the same single *** optimized RuGST electronic synapse with the high transformation temperature of hexagonal phase>380C,the outstanding endurance>108 cycles,the low resistance drift factor of 0.092,as well as the extremely high linearity with correlation coefficients of 0.999 and 0.976 in parts of potentiation and *** investigations also go insight to mechanisms of Ru doping according to thorough microstructure characterization,revealing that Ru dopant is able to enter GST lattices thus changing and stabilizing atomic arrangement of *** leads to the short-term plasticity realized by RuGST ***,the proposed RuGST electronic synapses performs a high accuracy of94.1%in a task of image recognition of CIFAR-100 database using ResNet *** work promotes the development of PCRAM platforms for large-scale neuromorphic systems.

关键词： electronic synapses neuromorphic computing PCRAM Ru-doped Ge_(2)Sb_(2)Te_(5)

DualDiff+: Dual-Branch Diffusion for High-Fidelity Video Generation with Reward Guidance

学校读者我要写书评

暂无评论

arXiv 2025年

作者： Yang, Zhao Qian, Zezhong Li, Xiaofan Xu, Weixiang Zhao, Gongpeng Yu, Ruohong Zhu, Lingsi Liu, Longjun The National Key Laboratory of Human-Machine Hybrid Augmented Intelligence National Engineering Research Center for Visual Information and Applications Institute of Artificial Intelligence and Robotics Xi’an Jiaotong University China The College of Optical Science and Engineering Zhejiang University Hangzhou310027 China Institute of Automation Chinese Academy of Sciences Beijing100190 China The University of Science and Technology of China Anhui230052 China

Accurate and high-fidelity driving scene reconstruction demands the effective utilization of comprehensive scene information as conditional inputs. Existing methods predominantly rely on 3D bounding boxes and BEV road maps for foreground and background control, which fail to capture the full complexity of driving scenes and adequately integrate multimodal information. In this work, we present DualDiff, a dual-branch conditional diffusion model designed to enhance driving scene generation across multiple views and video sequences. Specifically, we introduce Occupancy Ray-shape Sampling (ORS) as a conditional input, offering rich foreground and background semantics alongside 3D spatial geometry to precisely control the generation of both elements. To improve the synthesis of fine-grained foreground objects, particularly complex and distant ones, we propose a Foreground-Aware Mask (FGM) denoising loss function. Additionally, we develop the Semantic Fusion Attention (SFA) mechanism to dynamically prioritize relevant information and suppress noise, enabling more effective multimodal fusion. Finally, to ensure high-quality image-to-video generation, we introduce the Reward-Guided Diffusion (RGD) framework, which maintains global consistency and semantic coherence in generated videos. Extensive experiments demonstrate that DualDiff achieves state-of-the-art (SOTA) performance across multiple datasets. On the NuScenes dataset, DualDiff reduces the FID score by 4.09% compared to the best baseline. In downstream tasks, such as BEV segmentation, our method improves vehicle mIoU by 4.50% and road mIoU by 1.70%, while in BEV 3D object detection, the foreground mAP increases by 1.46%. Code will be made available at https://***/yangzhaojason/DualDiff. Copyright © 2025, The Authors. All rights reserved.

关键词： Object detection

Learnable Frequency Filters and Low-Frequency Domain Randomization Based Domain Generalization for Medical Image Segmentation

学校读者我要写书评

暂无评论

Learnable Frequency Filters and Low-Frequency Domain Randomi...

Chinese Automation Congress (CAC)

作者： Zikai Chen Jiayi Wu Jincheng Li Weiliang Zuo Chenzdong Li Jingmin Xin Nanning Zheng National Key Laboratory of Human-Machine Hybrid Augmented Intelligence National Engineering Research Center of Visual Information and Applications Institute of Artificial Intelligence and Robotics Xi'an Jiaotong University Xi'an China The First Affiliated Hospital College of Clinical Medicine Medical College of Henan University of Science and Technology Luoyang China

In recent years, the convolutional neural network represented by U-net has achieved great success in the field of medical image segmentation. However, due to different data acquisition protocols and differences in instrument parameters, the data in actual clinical applications will have a distribution shift problem with the training data of the model, also known as the domain shift problem. This can lead to a decrease in model performance, hindering the application of the model in real-world clinical settings. Traditional convolutional neural networks have local perception properties, and their inability to model long-range dependencies hinders the improvement of domain generalization performance. In addition, the traditional DG method of feature space domain randomization is not effective in medical image segmentation tasks due to the limited search space of feature styles and the inability to keep semantic information unchanged. To address the above issues, this paper proposes a medical image segmentation network named GFDR-U net for domain generalization. The network utilizes globally learnable filters to model long-range dependencies, and introduces a low-frequency domain randomization method to solve the problem of semantic invariance. Extensive experiments show that our method outperforms the state-of-the-art methods.

关键词：

Neural P3M: A Long-Range Interaction Modeling Enhancer for Geometric GNNs

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Wang, Yusong Cheng, Chaoran Li, Shaoning Ren, Yuxuan Shao, Bin Liu, Ge Heng, Pheng-Ann Zheng, Nanning National Key Laboratory of Human-Machine Hybrid Augmented Intelligence National Engineering Research Center for Visual Information and Applications Institute of Artificial Intelligence and Robotics Xi’an Jiaotong University China University of Illinois Urbana-Champaign United States Department of Computer Science and Engineering The Chinese University of Hong Kong Hong Kong University of Science and Technology of China China Microsoft Research AI4Science

关键词： Graph neural networks

Revisiting the Adversarial Robustness of Vision Language Models: a Multimodal Perspective

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Zhou, Wanqi Bai, Shuanghao Mandic, Danilo P. Zhao, Qibin Chen, Badong The National Key Laboratory of Human-Machine Hybrid Augmented Intelligence National Engineering Research Center for Visual Information and Applications Institute of Artificial Intelligence and Robotics Xi’an Jiaotong University Xi’an710049 China RIKEN AIP Tokyo Japan Guangdong University of Technology Guangdong510006 China The Department of Electrical Engineering Imperial College London LondonSW7 2BX United Kingdom

Pretrained vision-language models (VLMs) like CLIP exhibit exceptional generalization across diverse downstream tasks. While recent studies reveal their vulnerability to adversarial attacks, research to date has primarily focused on enhancing the robustness of image encoders against image-based attacks, with defenses against text-based and multimodal attacks remaining largely unexplored. To this end, this work presents the first comprehensive study on improving the adversarial robustness of VLMs against attacks targeting image, text, and multimodal inputs. This is achieved by proposing multimodal contrastive adversarial training (MMCoA). Such an approach strengthens the robustness of both image and text encoders by aligning the clean text embeddings with adversarial image embeddings, and adversarial text embeddings with clean image embeddings. The robustness of the proposed MMCoA is examined against existing defense methods over image, text, and multimodal attacks on the CLIP model. Extensive experiments on 15 datasets across two tasks reveal the characteristics of different adversarial defense methods under distinct distribution shifts and dataset complexities across the three attack types. This paves the way for a unified framework of adversarial robustness against different modality attacks, opening up new possibilities for securing VLMs against multimodal attacks. The code is available at https://***/ElleZWQ/***. Copyright © 2024, The Authors. All rights reserved.

关键词： visual languages

ERetinex: Event Camera Meets Retinex Theory for Low-Light Image Enhancement

学校读者我要写书评

暂无评论

arXiv 2025年

作者： Guo, Xuejian Tian, Zhiqiang Wang, Yuehang Li, Siqi Jiang, Yu Du, Shaoyi Gao, Yue the School of Software Engineering Xi’an Jiaotong University Xi’an710049 China the College of Computer Science and Technology Jilin University Changchun130012 China the National Key Laboratory of Human-Machine Hybrid Augmented Intelligence National Engineering Research Center for Visual Information and Applications Institute of Artificial Intelligence and Robotics Xi’an Jiaotong University Xi’an710049 China BNRist THUIBCS KLISS BLBCI School of Software Tsinghua University Beijing100084 China

Low-light image enhancement aims to restore the under-exposure image captured in dark scenarios. Under such scenarios, traditional frame-based cameras may fail to capture the structure and color information due to the exposure time limitation. Event cameras are bio-inspired vision sensors that respond to pixel-wise brightness changes asynchronously. Event cameras’ high dynamic range is pivotal for visual perception in extreme low-light scenarios, surpassing traditional cameras and enabling applications in challenging dark environments. In this paper, inspired by the success of the retinex theory for traditional frame-based low-light image restoration, we introduce the first methods that combine the retinex theory with event cameras and propose a novel retinex-based low-light image restoration framework named ERetinex. Among our contributions, the first is developing a new approach that leverages the high temporal resolution data from event cameras with traditional image information to estimate scene illumination accurately. This method outperforms traditional image-only techniques, especially in low-light environments, by providing more precise lighting information. Additionally, we propose an effective fusion strategy that combines the high dynamic range data from event cameras with the color information of traditional images to enhance image quality. Through this fusion, we can generate clearer and more detail-rich images, maintaining the integrity of visual information even under extreme lighting conditions. The experimental results indicate that our proposed method outperforms state-of-the-art (SOTA) methods, achieving a gain of 1.0613 dB in PSNR while reducing FLOPS by 84.28%. The code is available at https://***/lodew920/ERetinex. © 2025, CC BY.

关键词： Cameras

DVT-Tree: Dynamic Visible Topology Tree for Efficient Mapless Navigation in Maze Environments

学校读者我要写书评

暂无评论

DVT-Tree: Dynamic Visible Topology Tree for Efficient Maples...

International Conference on Intelligent Transportation

作者： Yucong Zhong Jianyi Liu Zhiqiang Jian Jinlong Pang Shitao Chen Xinhu Zheng National Key Laboratory of Human-Machine Hybrid Augmented Intelligence National Engineering Research Center of Visual Information and Applications Institute of Artificial Intelligence and Robotics Xi'an Jiaotong University Shaanxi P.R. China Shunan Academy of Artificial Intelligence Ningbo Zhejiang P.R. China Intelligent Transportation Thrust Hong Kong University of Science and Technology Guangzhou China

Path planning for autonomous mobile robots in unknown environments remains a challenging task, as it requires the ability to collect and process environmental information in real time for the exploration of uncertain areas. Currently, the state-of-the-art method effectively navigates by building a Rapid Visible Tree(RVT) [1] to reserve potential openings detected during exploration, thereby avoiding getting trapped in local minima and achieving good results. However, the RVT method regards the openings as one-dimensional frontiers between known and unknown areas without considering the spatial continuity, resulting in excessive redundancy of frontiers information in the tree. In this paper, we propose a novel, efficient and safe navigation method specifically tailored for scenarios where no prior maps are available. Technically, our method extracts the topological representation of the perception space, and selects topological zones for safe exploration. The robot is steered to discover the frontiers of the unknown environment, and the exploration strategy gets updated in real time as the environment evolves. As part of the exploration procedure, the Dynamic Visible Topology Tree (DVT-Tree) continually expands in response to the environment's topological data. The proposed method is validated on a two-wheeled differential robot platform in simulation as well as real-world. Comparative results with other existing methods confirm that our method outperforms in terms of efficiency and real-time response for navigation without a map in maze environments.

关键词：

Multi-risk Aware Trajectory Planning for Car-like Robot in Highly Dynamic Environments

学校读者我要写书评

暂无评论

Multi-risk Aware Trajectory Planning for Car-like Robot in H...

International Conference on Intelligent Transportation

作者： Ming Chen Jianyi Liu Jinlong Pang Zhiqiang Jian Pei Chen Xinhu Zheng National Key Laboratory of Human-Machine Hybrid Augmented Intelligence National Engineering Research Center of Visual Information and Applications Institute of Artificial Intelligence and Robotics Xi'an Jiaotong University Shaanxi P.R. China Shunan Academy of Artificial Intelligence Ningbo Zhejiang P.R. China Intelligent Transportation Thrust Hong Kong University of Science and Technology Guangzhou China

Safe trajectory planning in highly dynamic environments remains a substantial challenge. Traditional risk-based trajectory planning algorithms solve planning problems in environments where humans and robots coexist, by taking into account the risk of collision with both dynamic and static obstacles. However, in highly dynamic environments, their safety during navigation is not guaranteed due to the singularity of risk assessment. Therefore, we first introduce a time-varying density risk map to represent external risks, then introduce a novel interact risk map generated based on the relative state of the robot and dynamic obstacles to represent interact risks. Finally, these two risks are combined based on a convex combination model to obtain a fusion risk map. The planner searches for an efficient, kinematically feasible, and safe trajectory by exploring the state graph built on motion primitives based on the fusion risk map. We have conducted a large number of simulation experiments in various challenging environments, and the experimental results have verified the effectiveness of the algorithm.

关键词：