检索结果-内蒙古大学图书馆

Operation-Level Performance Benchmarking of Graph Neural Networks for Scientific Applications

学校读者我要写书评

暂无评论

arXiv 2022年

作者： Hosseini, Ryien Simini, Filippo Vishwanath, Venkatram Argonne Leadership Computing Facility Argonne National Laboratory Lemont IL United States Department of Electrical Engineering and Computer Science University of Michigan Ann ArborMI United States

As Graph Neural Networks (GNNs) increase in popularity for scientific machine learning, their training and inference efficiency is becoming increasingly critical. Additionally, the deep learning field as a whole is trending towards wider and deeper networks, and ever increasing data sizes, to the point where hard hardware bottlenecks are often encountered. Emerging specialty hardware platforms provide an exciting solution to this problem. In this paper, we systematically profile and select low-level operations pertinent to GNNs for scientific computing implemented in the Pytorch Geometric software framework. These are then rigorously benchmarked on NVIDIA A100 GPUs for several various combinations of input values, including tensor sparsity. We then analyze these results for each operation. At a high level, we conclude that on NVIDIA systems: (1) confounding bottlenecks such as memory inefficiency often dominate runtime costs moreso than data sparsity alone, (2) native Pytorch operations are often as or more competitive than their Pytorch Geometric equivalents, especially at low to moderate levels of input data sparsity, and (3) many operations central to state-of-the-art GNN architectures have little to no optimization for sparsity. We hope that these results serve as a baseline for those developing these operations on specialized hardware and that our subsequent analysis helps to facilitate future software and hardware based optimizations of these operations and thus scalable GNN performance as a whole. © 2022, CC BY.

关键词： Graph neural networks

Deep imaging inside scattering media through virtual spatiotemporal wavefront shaping

学校读者我要写书评

暂无评论

arXiv 2023年

作者： Zhang, Yiwen Dinh, Minh Wang, Zeyu Zhang, Tianhao Chen, Tianhang Hsu, Chia Wei Ming Hsieh Department of Electrical and Computer Engineering University of Southern California Los AngelesCA90089 United States Interdisciplinary Center for Quantum Information State Key Laboratory of Modern Optical Instrumentation College of Information Science and Electronic Engineering Zhejiang University Hangzhou310027 China

The multiple scattering of light makes materials opaque and obstructs imaging. Wavefront shaping can reverse the scattering process, but imaging with physical wavefront shaping has limitations such as requiring physical guidestars, being restricted within a small isoplanatic volume, and relying on slow wavefront updates, with some approaches only working for planar targets outside the scattering media. Here, we introduce scattering matrix tomography (SMT): measure the hyperspectral scattering matrix of the sample, use it to digitally scan a synthesized confocal spatiotemporal focus and construct a volumetric image of the sample, and then use the tomograms as virtual guidestars in a nonconvex optimization to find the pulse shape, input wavefront, and output wavefront that can compensate for aberrations and scattering. SMT combines the strengths of wavefront shaping, spatiotemporal gating, and computational adaptive optics, eliminating physical guidestars and enabling digital double-path wavefront corrections tailored for every isoplanatic volume. We demonstrate sub-micron lateral resolution and one-micron axial resolution at one millimeter beneath ex vivo mouse brain tissue and over three transport mean free paths inside an opaque colloid, where existing imaging methods all fail due to the overwhelming multiple scattering. As a noninvasive and label-free method that works both inside and outside the scattering media, SMT may be applied broadly across medical imaging, biological science, device inspection, and colloidal physics. Copyright © 2023, The Authors. All rights reserved.

关键词： Wavefronts

CLDG: Contrastive Learning on Dynamic Graphs

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Xu, Yiming Shi, Bin Ma, Teng Dong, Bo Zhou, Haoyi Zheng, Qinghua Department of Computer Science and Technology Xi’an Jiaotong University China Shaanxi Provincial Key Laboratory of Big Data Knowledge Engineering Xi’an Jiaotong University China Department of Distance Education Xi’an Jiaotong University China School of Software Beihang University China Advanced Innovation Center for Big Data and Brain Computing Beihang University China

The graph with complex annotations is the most potent data type, whose constantly evolving motivates further exploration of the unsupervised dynamic graph representation. One of the representative paradigms is graph contrastive learning. It constructs self-supervised signals by maximizing the mutual information between the statistic graph’s augmentation views. However, the semantics and labels may change within the augmentation process, causing a significant performance drop in downstream tasks. This drawback becomes greatly magnified on dynamic graphs. To address this problem, we designed a simple yet effective framework named CLDG. Firstly, we elaborate that dynamic graphs have temporal translation invariance at different levels. Then, we proposed a sampling layer to extract the temporally-persistent signals. It will encourage the node to maintain consistent local and global representations, i.e., temporal translation invariance under the timespan views. The extensive experiments demonstrate the effectiveness and efficiency of the method on seven datasets by outperforming eight unsupervised state-of-the-art baselines and showing competitiveness against four semi-supervised methods. Compared with the existing dynamic graph method, the number of model parameters and training time is reduced by an average of 2,001.86 times and 130.31 times on seven datasets, respectively. The code and data are available at: https://***/yimingxu24/CLDG. © 2024, CC BY.

关键词： Contrastive Learning

Reinforcement Learning with Unbiased Policy Evaluation and Linear Function Approximation

学校读者我要写书评

暂无评论

arXiv 2022年

作者： Winnicki, Anna Srikant, R. Coordinated Science Laboratory The Department of Electrical and Computer Engineering University of Illinois Urbana-Champaign United States C3.ai Digital Transformation Institute United States

We provide performance guarantees for a variant of simulation-based policy iteration for controlling Markov decision processes that involves the use of stochastic approximation algorithms along with state-of-the-art techniques that are useful for very large MDPs, including lookahead, function approximation, and gradient descent. Specifically, we analyze two algorithms;the first algorithm involves a least squares approach where a new set of weights associated with feature vectors is obtained via least squares minimization at each iteration and the second algorithm involves a two-time-scale stochastic approximation algorithm taking several steps of gradient descent towards the least squares solution before obtaining the next iterate using a stochastic approximation algorithm. Copyright © 2022, The Authors. All rights reserved.

关键词： Markov processes

Wheat Rust Disease Segmentation from Ground Imagery

学校读者我要写书评

暂无评论

Wheat Rust Disease Segmentation from Ground Imagery

Proceedings of the Digital Image Computing: Technqiues and Applications (DICTA)

作者： Hirra Anwar Muhammad Jawad Khan Muhammad Fayyaz Ajmal Saeed Mian Faisal Shafait School of Mechanical and Manufacturing Engineering NUST Islamabad Pakistan Department of Electrical Engineering Prince Sattam Bin Abdul Aziz University Al-Kharj Saudi Arabia Crop Diseases Research Institute National Agriculture Research Center Islamabad Pakistan University of Western Australia Perth Australia School of Electrical Engineering & Computer Science NUST Islamabad Pakistan Deep Learning Laboratory National Center of Artificial Intelligence Islamabad Pakistan

ISBN: (数字)9798350379037

ISBN: (纸本)9798350379044

Wheat rust disease is a major agricultural threat, causing significant yield losses globally. Precision agriculture and autonomous robot systems represent pivotal advancements in modern farming and disease detection, leveraging technology to enhance crop management, optimize resource use, and improve productivity. These systems rely heavily on accurate and comprehensive data to train effectively. However, current datasets often fall short of meeting the specific needs of these advanced agricultural technologies. This paper addresses this gap by introducing a primary contribution in the form of an open-source dataset focused on wheat rust disease segmentation for precision agriculture. Open-source datasets in literature are either general plant disease datasets or wheat disease classification datasets. Our proposed wheat rust segmentation dataset, Focussed Rust Disease Images (FRDI), provides close-up views of disease manifestations on wheat leaves, capturing the natural variability and complexity of diseases in field conditions, thereby meeting the requirements of precision agriculture systems. This facilitates detailed analysis and effective training of deep learning models designed for disease detection and segmentation tasks for real-world agricultural environments. Furthermore, in our work we present an improved wheat rust disease detection and segmentation technique which is based on the concept of co-salient feature extraction for disease identification within complex and cluttered field images. By integrating localized and generic contextual information, our proposed approach effectively identifies key features and intricate patterns associated with wheat rust disease.

关键词： Precision agriculture Training Deep learning Representation learning Productivity Image segmentation Plant diseases Object detection Feature extraction Diseases

Tunable Magnetically Induced Transparency Spectra in Magnon-Magnon Coupled Y3Fe5O12/Permalloy Bilayers

学校读者我要写书评

暂无评论

Physical Review Applied 2022年第4期17卷 044010-044010页

作者： Yuzan Xiong Jerad Inman Zhengyi Li Kaile Xie Rao Bidthanapally Joseph Sklenar Peng Li Steven Louis Vasyl Tyberkevych Hongwei Qu Zhili Xiao Wai K. Kwok Valentine Novosad Yi Li Fusheng Ma Wei Zhang Department of Physics Oakland University Rochester Michigan 48309 USA Department of Electronic and Computer Engineering Oakland University Rochester Michigan 48309 USA Materials Science Division Argonne National Laboratory Argonne Illinois 60439 USA Jiangsu Key Laboratory of Opto-Electronic Technology School of Physics and Technology Nanjing Normal University Nanjing 210046 China Department of Physics and Astronomy Wayne State University Detroit Michigan 48201 USA Department of Electrical and Computer Engineering Auburn University Auburn Alabama 36849 USA

Hybrid magnonic systems host a variety of characteristic phenomena such as the magnetically induced transparency (MIT) and Purcell effect, which are considered useful for future coherent quantum-information processing. In this work, we experimentally demonstrate a tunable MIT effect in the Y3Fe5O12(YIG)/Permalloy(Py) magnon-magnon coupled system via changing the magnetic field orientations. By probing the magneto-optic effects of Py and YIG thin films, we identify clear features of MIT spectra induced by the mode hybridization between the uniform mode of Py and the perpendicular standing spin-wave modes of YIG. By changing the external magnetic field orientations, we observe a tunable coupling strength between the YIG’s spin-wave modes and the Py’s uniform mode, upon the application of an out-of-plane magnetic field. This observation is theoretically interpreted by a geometrical consideration of the Py and YIG magnetization under the oblique magnetic field even at a constant interfacial exchange coupling. Our findings show high promise for investigating tunable coherent phenomena with hybrid magnonic platforms.

关键词： Electromagnetically induced transparency Magneto-optical effect Magnons Quantum information with hybrid systems Spintronics Ferromagnets Magnetic insulators Magnetic multilayers

Global Stability and Optimal Control of Epidemics in Heterogeneously Structured Populations Exhibiting Adaptive Behaviour

学校读者我要写书评

暂无评论

SSRN

SSRN 2023年

作者： Liu, Lixiang Chen, Shanshan Small, Michael Moore, Jack Murdoch Shang, Keke Department of Computer Science School of Electronic and Electrical Engineering Shanghai University of Engineering Science Shanghai201620 China Complex Systems Group Department of Mathematics and Statistics The University of Western Australia Perth6009 Australia Mineral Resources Commonwealth Scientific and Industrial Research Organisation Kensington6151 Australia MOE Key Laboratory of Advanced Micro-Structured Materials School of Physics Science and Engineering Tongji University Shanghai200092 China Computational Communication Collaboratory Nanjing University Nanjing210093 China

During the outbreak of infectious diseases, individual always make some adaptive response to avoid infected the diseases. In this paper, we introduce SIRS model on scale-free networks that take into account behavioral memory and time delay to depict individuals’ adaptive feedback, which is better characterize the actual spread of epidemics. We address uniform persistence and find that the model has the sharp threshold property, that is, for the basic reproduction number less than one, the disease-free equilibrium is globally asymptotically stable, while for the basic reproduction number is above one, the endemic equilibrium is globally stable. We further analyze the effects of two control strategies (the vaccination of susceptible people and treatment of infected people) on epidemic dynamics by optimal control strategy. The existence and uniqueness of the optimal control strategy are obtained. Finally, we perform some stochastic network simulations that yield quantitative agreement with the deterministic mean-field approach. The final results show that the effect of time delay does not change the basic reproduction number, but it plays an important role in the process of disease transmission. The behavioral memory is able to affect the basic reproduction number, and it has important and positive implications for disease transmission and control. The work of this paper can provide a theoretical basis and reference for the prediction and prevention of infectious diseases. © 2023, The Authors. All rights reserved.

关键词： Diseases

Shear strain alters the structure and migration mechanism of self-interstitial atoms in copper

学校读者我要写书评

暂无评论

Physical Review Materials 2022年第5期6卷 053605-053605页

作者： B. Zhang C. Wheatley P. Chen X. Qian M. J. Demkowicz Department of Materials Science and Engineering Texas A&M University College Station Texas 77845 USA Department of Materials Science and Engineering Penn State University State College Pennsylvania 16801 USA Key Laboratory of Automobile Materials of Ministry of Education School of Materials Science and Engineering Jilin University Changchun People's Republic of China Department of Electrical and Computer Engineering Texas A&M University College Station Texas 77845 USA

We use atomistic modeling to show that externally applied shear strain causes the lowest energy self-interstitial atom (SIA) structure in copper (Cu) to change from a ⟨100⟩-type dumbbell to a ⟨110⟩-type dumbbell. Concurrently, SIA migration switches from the three-dimensional (3D) random walk characteristic of ⟨100⟩-type dumbbells to a 1D mechanism analogous to that of crowdion SIAs. Furthermore, the relative energies of these two dumbbell structures as a function of strain are well predicted using elastic dipole tensors computed at zero strain, indicating that examination of these tensors may be used to assess the likelihood of strain-induced SIA structure transitions in other materials. Changes in lowest energy SIA structures and associated migration mechanisms stand to impact predictions of SIA behavior in irradiated solids.

关键词： Interstitials Face-centered cubic Transition metals Density functional calculations

Unification of Finite Symmetries in Simulation of Many-body Systems on Quantum computers

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Bastidas, Victor M. Fitzpatrick, Nathan Joven, K.J. Rossi, Zane M. Islam, Shariful Van Voorhis, Troy Chuang, Isaac L. Liu, Yuan Department of Chemistry Massachusetts Institute of Technology CambridgeMA02139 United States Physics and Informatics Laboratory NTT Research Inc. 940 Stewart Dr. SunnyvaleCA94085 United States Quantinuum Terrington House 13-15 Hills Road CambridgeCB2 1NL United Kingdom Department of Physics Massachusetts Institute of Technology CambridgeMA02139 United States Department of Physics North Carolina State University RaleighNC27606 United States Department of Electrical Engineering and Computer Science Massachusetts Institute of Technology CambridgeMA02139 United States Department of Electrical and Computer Engineering North Carolina State University RaleighNC27606 United States Department of Computer Science North Carolina State University RaleighNC27606 United States

Symmetry is fundamental in the description and simulation of quantum systems. Leveraging symmetries in classical simulations of many-body quantum systems often results in an exponential overhead due to the exponentially growing size of some symmetry groups as the number of particles increases. Quantum computers hold the promise of achieving exponential speedup in simulating quantum many-body systems;however, a general method for utilizing symmetries in quantum simulations has not yet been established. In this work, we present a unified framework for incorporating symmetry groups into the simulation of many-body systems on quantum computers. The core of our approach lies in the development of efficient quantum circuits for symmetry-adapted projection onto irreducible representations of a group or pairs of commuting groups. We provide resource estimations for common groups, including the cyclic and permutation groups. Our algorithms demonstrate the capability to prepare coherent superpositions of symmetry-adapted states and to perform quantum evolution across a wide range of models in condensed matter physics and ab initio electronic structure in quantum chemistry. Specifically, we execute a symmetry-adapted quantum subroutine for small molecules in first quantization on noisy hardware, and demonstrate the emulation of symmetry-adapted quantum phase estimation for preparing coherent superpositions of quantum states in various irreps of a symmetry group. In addition, we present a discussion of major open problems regarding the use of symmetries in digital quantum simulations of many-body systems, paving the way for future systematic investigations into leveraging symmetries for practical quantum advantage. The broad applicability and rigorous resource estimation for symmetry transformations make our framework appealing for achieving provable quantum advantage on fault-tolerant quantum computers. The efficiency of the proposed symmetry-adapted subroutine holds the promi

关键词： Quantum computers