检索结果-内蒙古大学图书馆

IEEE/ACM workshop on machine learning in high performance computing environments (mlhpc) / workshop on Artificial Intelligence and machine learning for Scientific Applications (AI4S)

作者： Botelho, Sergio Joshi, Ameya Khara, Biswajit Rao, Vinay Sarkar, Soumik Hegde, Chinmay Adavani, Santi Ganapathysubramanian, Baskar RocketML Inc Beaverton OR 97005 USA NYU New York NY 10003 USA Iowa State Univ Ames IA USA

ISBN: (纸本)9781665422918

Recent progress in scientific machine learning (SciML) has opened up the possibility of training novel neural network architectures that solve complex partial differential equations (PDEs). Several (nearly data free) approaches have been recently reported that successfully solve PDEs, with examples including deep feed forward networks, generative networks, and deep encoder-decoder networks. However, practical adoption of these approaches is limited by the difficulty in training these models, especially to make predictions at large output resolutions (>= 1024 x 1024). Here we report on a software framework for data parallel distributed deep learning that resolves the twin challenges of training these large SciML models training in reasonable time as well as distributing the storage requirements. Our framework provides several out of the box functionality including (a) loss integrity independent of number of processes, (b) synchronized batch normalization, and (c) distributed higher-order optimization methods. We show excellent scalability of this framework on both cloud as well as HPC clusters, and report on the interplay between bandwidth, network topology and bare metal vs cloud. We deploy this approach to train generative models of sizes hitherto not possible, showing that neural PDE solvers can be viably trained for practical applications. We also demonstrate that distributed higher-order optimization methods are 2-3 x faster than stochastic gradient-based methods and provide minimal convergence drift with higher batch-size.

关键词： Deep generative models Distributed training PDEs Loss functions Cloud vs HPC higher-order optimization

来源：评论

学校读者我要写书评

暂无评论

Comparative Study of Embedded Energy Management Methods based on machine learning for Dual-Source Electric Vehicles

引用

IEEE Transactions on Transportation Electrification 2025年

作者： Adnane, Marouane Nguyen, Bao-Huy Khoumsi, Ahmed Trovao, Joao Pedro F. Université de Sherbrooke Department of Electrical Engineering and Computer Engineering SherbrookeQCJ1K 2R1 Canada Hanoi University of Science and Technology CTI Lab4EV School of Electrical and Electronic Engineering Hanoi Viet Nam

In the actual context of dual-source electric vehicles (DSEVs), efficient energy management strategies (EMSs) are essential to optimize energy distribution between batteries and supercapacitors. However, achieving real-time decision-making and adaptability in resource-constrained environments remains a significant challenge. This study addresses these challenges by comparing two advanced machine learning (ML)-based EMSs: Driving Mode Prediction-based EMS (DMP-EMS) and Multi-Ensemble learning-based EMS (MEL-EMS). Using Signal Hardware-In-the-Loop simulation and a Raspberry Pi 4 platform, the two EMSs are evaluated based on various criteria, including performance accuracy, model complexity, training and prediction times, required computing resources, scalability, and energy consumption. The findings highlight the trade-offs between these EMSs, offering insights into designing tailored EMSs for DSEVs under varying operational demands. While MEL-EMS excels in environments where prediction accuracy is paramount since it offers superior predictive performance suitable for applications demanding high accuracy and scalability, DMP-EMS provides a more efficient and practical solution for real-time energy management in resource-constrained settings. © 2015 IEEE.

关键词： Hardware-in-the-loop simulation

来源：评论

学校读者我要写书评

暂无评论

Proceedings of PEHC 2021: workshop on Programming environments for Heterogeneous computing, Held in conjunction with SC 2021: The International Conference for high performance computing, Networking, Storage and Analysis

Proceedings of PEHC 2021: Workshop on Programming Environmen...

引用

2021 IEEE/ACM workshop on Programming environments for Heterogeneous computing, PEHC 2021

ISBN: (纸本)9781728186764

The proceedings contain 6 papers. The topics discussed include: GenMAT: a general-purpose machine learning-driven auto-tuner for heterogeneous platforms;OSCAR parallelizing and power reducing compiler and API for heterogeneous multicores;a python-based high-level programming flow for CPU-FPGA heterogeneous systems;a holistic systems approach to leveraging heterogeneity;survival of the fittest amidst the Cambrian explosion of processor architectures for artificial intelligence;and designing heterogeneous systems: large scale architectural exploration via simulation.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Deep learning-based Low-dose Tomography Reconstruction with Hybrid-dose Measurements 6

Deep Learning-based Low-dose Tomography Reconstruction with ...

引用

IEEE/ACM workshop on machine learning in high performance computing environments (mlhpc) / workshop on Artificial Intelligence and machine learning for Scientific Applications (AI4S)

作者： Wu, Ziling Bicer, Tekin Liu, Zhengchun De Andrade, Vincent Zhu, Yunhui Foster, Ian T. Virginia Polytech Inst & State Univ Bradley Dept Elect & Comp Engn Blacksburg VA 24061 USA Argonne Natl Lab Data Sci & Learning Div Argonne IL 60439 USA Argonne Natl Lab Xray Sci Div Argonne IL 60439 USA

ISBN: (纸本)9781665422918

Synchrotron-based X-ray computed tomography is widely used for investigating inner structures of specimens at high spatial resolutions. However, potential beam damage to samples often limits the X-ray exposure during tomography experiments. Proposed strategies for eliminating beam damage also decrease reconstruction quality. Here we present a deep learning-based method to enhance low-dose tomography reconstruction via a hybrid-dose acquisition strategy composed of extremely sparse-view normal-dose projections and full-view low-dose projections. Corresponding image pairs are extracted from low-/normal-dose projections to train a deep convolutional neural network, which is then applied to enhance full-view noisy low-dose projections. Evaluation on two experimental datasets under different hybrid-dose acquisition conditions show significantly improved structural details and reduced noise levels compared to uniformly distributed acquisitions with the same number of total dosage. The resulting reconstructions also preserve more structural information than reconstructions processed with traditional analytical and regularization-based iterative reconstruction methods from uniform acquisitions. Our performance comparisons show that our implementation, HDrec, can perform denoising of a real-world experimental data 410x faster than the state-of-the-art Xlearn method while providing better quality. This framework can be applied to other tomographic or scanning based X-ray imaging techniques for enhanced analysis of dose-sensitive samples and has great potential for studying fast dynamic processes.

关键词： low-dose tomography image reconstruction hybrid-dose measurement projection denoising deep learning

来源：评论

学校读者我要写书评

暂无评论

Architectural Requirements for Deep learning Workloads in HPC environments 12

Architectural Requirements for Deep Learning Workloads in HP...

引用

12th International workshop on performance Modeling, Benchmarking and Simulation of high performance Computer Systems held as part of the 33rd ACM/IEEE International Conference for high performance computing, Networking, Storage and Analysis

作者： Ibrahim, Khaled Z. Tan Nguyen Hai Ah Nam Bhimji, Wahid Farrell, Steven Oliker, Leonid Rowan, Michael Wright, Nicholas J. Williams, Samuel Lawrence Berkeley Natl Lab NERSC CRD Berkeley CA 94720 USA

ISBN: (纸本)9781665411189

Scientific machine learning (SciML) promises to have a transformational impact on scientific exploration, by combining state-of-the-art AI methods with the latest generation of supercomputers. However, to efficiently leverage ML techniques on high-performance computing (HPC) systems, it is critical to understand the performance characteristics of the underlying algorithms on modern computational systems. In this work, we present a new methodology for developing a detailed performance understanding of ML benchmarks. To demonstrate our approach we investigate two emerging SciML benchmark applications from cosmology and climate, ComsoFlow and DeepCAM, as well as ResNet-50, a well-known image classification model. We develop and validate performance models that explore the key architectural artifacts, including memory requirements, data reuse, and performance efficiency across both single- and multiple-GPU computations. Our methodology also focuses on the complexity of data-movement across storage and memory hierarchies, and leverages our performance models to capture key components of runtime execution while highlighting design tradeoffs. Although our work focuses on image-processing methods on GPU-based HPC systems, our approach is applicable to a variety of ML algorithmic domains and emerging AI accelerators. Overall, our insights will help computer architects and data scientists understand performance bottlenecks and optimization opportunities to improve SciML design and system efficiency.

关键词： Runtime machine learning algorithms Computational modeling Memory management Benchmark testing Supercomputers Computational efficiency

来源：评论

学校读者我要写书评

暂无评论

AI-Driven Personalization in Organizational Communication

AI-Driven Personalization in Organizational Communication

引用

IEEE Communication Strategies in Digital Society workshop (ComSDS)

作者： Denis Shakhov Ruslan Batashev Ilyos Abdullayev Department of Economics and Management Khorezm University of Economics Urgench Uzbekistan Department of Taxes and Taxation Kadyrov Chechen State University Grozny Russia Department of Business and Management Urgench State University Urgench Uzbekistan

ISBN: (数字)9798331536121

ISBN: (纸本)9798331536138

This study investigates the transformative impact of artificial intelligence on corporate communications, focusing on AI-powered personalization systems in business environments. Through a systematic literature review (2019-2024), the research establishes an empirical framework for evaluating these systems’ technological infrastructure. The findings reveal distinct sector-specific performance variations: the retail sector showing 58% enhanced engagement metrics, while B2B segments demonstrated 28% improvement in key performance indicators. The technological foundation comprises machine learning algorithms, natural language processing frameworks, and high-performance computing systems enabling real-time personalization. The methodology integrates the adaptive personalization framework (APF) with the multidimensional personalization model (MPM) to elucidate machine learning mechanisms. This framework supports user profiling, navigation optimization, and behavioral pattern modification, secured through distributed ledger technologies. Empirical analysis reveals the complementarity between AI and human capabilities. While AI systems excel in response velocity (mean: 4.92), human interactions demonstrate superior responsiveness (5.27) and professional competency metrics (5.32 vs. 4.87), suggesting the optimality of a hybrid model. The study culminates in a conceptual framework balancing communication scalability with personalized relevance while adhering to ethical imperatives of data protection, algorithmic fairness, and transparency protocol.

关键词： Seminars Ethics Scalability Industrial communication machine learning Natural language processing Real-time systems Optimization Standards Systematic literature review

来源：评论

学校读者我要写书评

暂无评论

End-to-End Requirements Mapping for Cloud-Native Applications Using ML Models

End-to-End Requirements Mapping for Cloud-Native Application...

引用

IEEE Annual computing and Communication workshop and Conference (CCWC)

作者： Mbarka Soualhia Fetahi Wuhib Ericsson Research Montreal QC Canada

ISBN: (数字)9798331507695

ISBN: (纸本)9798331507701

In the rapidly evolving landscape of 5G and emerging 6G networks, ensuring cloud-based applications meet stringent performance and reliability requirements is critical to maintaining a competitive edge and satisfying user demands. These applications are typically composed of interconnected components, such as Virtual Network Functions (VNFs) or microservices, each contributing to the application's overall performance. However, configuring individual VNFs or microservices to meet specific end-to-end requirements—such as latency and throughput—poses a significant challenge. Current methods often rely on manual configurations, which can lead to inefficiencies and suboptimal resource use in dynamic cloud environments. In this paper, we introduce a machine learning-based solution to automate the mapping of end-to-end requirements into individual KPI requirements for each component (e.g., VNF) in the service graph. Using Deep Neural Networks (DNN) and Graph Neural Networks (GNN), our solution learns from diverse deployment scenarios and end-to-end application requirements, enabling accurate prediction of the KPIs needed for each component to support overall performance goals. This approach ensures that all components are precisely configured to meet end-to-end requirements without extensive manual tuning. Our evaluation on two distinct applications demonstrates high accuracy in predictions, achieving up to 98.99% accuracy during training and up to 96.67% in inferencing, surpassing the accuracy of expert-configured setups.

关键词： Training Cloud computing Accuracy Microservice architectures Manuals Artificial neural networks Dynamic scheduling Graph neural networks Reliability Tuning

来源：评论

学校读者我要写书评

暂无评论

workshop Organization – mlhpc 2020

Workshop Organization – MLHPC 2020

引用

IEEE/ACM workshop on machine learning in high performance computing environments (mlhpc)

Provides a listing of current committee members and society officers.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Message from the mlhpc workshop Chairs

Message from the MLHPC Workshop Chairs

引用

IEEE/ACM workshop on machine learning in high performance computing environments (mlhpc)

Presents the introductory welcome message from the conference proceedings. May include the conference officers' congratulations to all involved with the conference event and publication of the proceedings record.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Practical Federated learning Infrastructure for Privacy-Preserving Scientific computing

Practical Federated Learning Infrastructure for Privacy-Pres...

引用

Artificial Intelligence and machine learning for Scientific Applications (AI4S), IEEE/ACM International workshop on

作者： Lesi Wang Dongfang Zhao Department of Computer Science and Engineering University of Nevada Reno United States

ISBN: (纸本)9781665462082

Federated learning (FL) is deemed a promising paradigm for privacy-preserving data analytics in collaborative scientific computing. However, there lacks an effective and easy-to-use FL infrastructure for scientific computing and high-performance computing (HPC) environments. The objective of this position paper is two-fold. Firstly, we identify three missing pieces of a scientific FL infrastructure: (i) a native MPI programming interface that can be seamlessly integrated into existing scientific applications, (ii) an independent data layer for the FL system such that the user can pick the persistent medium for her own choice, such as parallel file systems and multidimensional databases, and (iii) efficient encryption protocols that are optimized for scientific workflows. The second objective of this paper is to present a work-in-progress FL infrastructure, namely MPI-FL, which is implemented with PyTorch and MPI4py. We deploy MPI-FL on 1,000 CPU cores and evaluate it with four standard benchmarks: MNIST, Fashion-MNIST, CIFAR-10, and SVHN-extra. It is our hope that the scientific computing and HPC community would find MPI-FL useful for their FL-related projects.

关键词： Protocols Data analysis Scientific computing Federated learning File systems Databases high performance computing

来源：评论

学校读者我要写书评

暂无评论

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案：

请选择收藏分类：

通借通还

建议与咨询 留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案： 新增检索档案 确定 取消

请选择收藏分类： 新增自定义分类 确定 取消

通借通还

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

请选择保存的检索档案：

请选择收藏分类：