检索结果-内蒙古大学图书馆

Using a multitasking GPU environment for content-based similarity measures of big data

Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) 2013年 7971卷 181-196页

作者： Tarakji, Ayman Hassani, Marwan Lankes, Stefan Seidl, thomas RWTH Aachen University Aachen Germany Data Management and Data Exploration Group RWTH Aachen University Germany

ISBN: (纸本)9783642396366

Performance and efficiency became recently key requirements of computer architectures. Modern computers incorporate Graphics processing Units (GPUs) into running data mining algorithms, as well as other general purpose computations. In this paper, different parallelization methods are analyzed and compared in order to understand their applicability. From multi-threading on shared memory to using NVIDIA's GPU accelerators for increasing performance and efficiency on parallel computing, this work discusses the parallelization of data mining algorithms considering performance and efficiency issues. the performance is compared on both many-core systems and GPU accelerators on a distance measure algorithm using a relatively big data set. We optimize the way we deal with GPUs in heterogeneous systems to make them more suitable for big data mining applications with heavy distance calculations. Moreover, we focus on achieving a higher utilization of GPU resources and a better reuse of data. Our implementation of the content-based similarity algorithm SQFD on the GPU outperforms by up to 50 x CPU counterparts, and up to 15x CPU multi-threaded implementations. © Springer-Verlag Berlin Heidelberg 2013.

关键词： Graphics processing unit

来源：评论

学校读者我要写书评

暂无评论

OpenMP-Based Monte Carlo Dose Calculation for Radiotherapy Treatment Planning on the Intel MIC Architecture

OpenMP-Based Monte Carlo Dose Calculation for Radiotherapy T...

引用

IEEE international Symposium on Information (IT) in Medicine and Education, ITME

作者： Qinglin Wang Jie Liu Peizhen Xie Chunye Gong Yuan Li Zuocheng Xing School of Computer Science National University of Defense Technology Changsha China Science and Technology on Parallel and Distributed Processing Laboratory National University of Defense Technology Changsha China

Monte Carlo (MC) simulation plays an important part in dose calculation for radiotherapy treatment planning. Since the accuracy of MC simulation relies on the number of simulated particles histories, it's very time-consuming. the Intel Many Integrated Core (MIC) architecture, which consists of more than 50 cores and supports many parallel programming models, provides an efficient alternative for accelerating MC dose calculation. this paper implements the OpenMP-based MC Dose Planning Method (DPM) for radiotherapy treatment problems on the Intel MIC architecture. the implementation has been verified on the target MIC coprocessor including 57 cores. the results demonstrate that the OpenMP-based DPM implementation exhibits very accurate results and achieves the maximum speedup of 10.53 times in comparison to the original DPM one on a Xeon E5-2670 CPU. Additionally, speedup and efficiency of the implementation running on the different number of cores in MIC are also reported.

关键词： Microwave integrated circuits Instruction sets Computer architecture Photonics Computational modeling History Interpolation

来源：评论

学校读者我要写书评

暂无评论

Rose Plant Disease Detection using Deep Learning

Rose Plant Disease Detection using Deep Learning

引用

international conference on Trends in Electronics and Informatics (ICEI)

作者： Md. Ali- Al - Alvy Golam Kibria Khan Mohammad Jahangir Alam Saiful Islam Mokhlesur Rahman Mirza Shahriyar Rahman Dept.of CSE Daffodil International University Dhaka Bangladesh

the detection and identification of rose plant disease is the focus of this investigation. Identification and detection are essential components of contemporary agro technology. In this case, AI technology was utilized to identify a disease in rose plants, although plant disease detection is difficult for sustainable agriculture. there are several instances of rose plant disease, and as a result, fascinating decoration is being lost. Due to this situation, which is getting worse every day in Bangladesh, the economy of agricultural sector is suffering. Bangladesh's population relies heavily on agriculture industry for their revenue. this study includes some disease detection of rose plants, albeit not all plants are affected equally by the illness. the plant leaf provides the plant with vital sustenance. When a leaf is ill, the plant is at its most vulnerable. Due to the accessibility of the sick leaf, disease identification is difficult. Agriculture field must be properly assessed to see significant improvements in proposed work. the best resource for creating this kind of disease detection model is deep learning technology. Image pre-processing and model analysis are steps in the disease detection construction process. Few CNN architectures are used in this study, including ResNet50, VGG-16 (Visual Geometry Group), MobileNetV2, and Inception V3. Four diseases have been identified in rose plant leaves. Here, image processing is investigated using a discovered approach and obtain a MobileNetV2 model accuracy of 96.11%.

关键词： Deep learning Plant diseases Visualization Plants (biology) Sociology Agriculture Internet of things

来源：评论

学校读者我要写书评

暂无评论

On the development of mg sheets with improved formability by applying additional shear strain during processing 7th

On the development of mg sheets with improved formability by...

引用

7th international Light Metals Technology conference, LMT 2015

作者： Letzig, D. Suh, J. Bohlen, J. Yi, S. Kurz, G. Golle, R. Volk, W. Victoria-Hernandez, J. Magnesium Innovation Centre Helmholtz Zentrum Geesthacht Max-Planck Strasse 1 Geesthacht21502 Germany Institute of Metal Forming and Casting Technical University of Munich Walther-Meissner Strasse 4 Garching85748 Germany

ISBN: (纸本)9783038355625

Reducing vehicle weight and emissions by lightweight design is a major goal of the automotive industry. Magnesium as the lightest structural metal offers a significant weight saving potential compared to steel and aluminum. However, the poor formability of magnesium semi-finished products (e.g. sheets) has hindered the massive application of this metal. this poor formability arises due to the formation of strong textures. Typically the basal planes of the HCP-structure align parallel to the sheet plane. Such preferred orientation of the basal planes limits the ability of basal slip to accommodate plastic strain in the sheet plane and is thus unfavourable for ductility and formability of the sheets, especially at room temperature. A combination of new technologies could help to alleviate the strong textures formed in Mg sheets. In this regard the utilization of twin-roll casting (TRC) in combination with processing techniques which apply additional shear strain such as differential speed rolling (DSR) and/or equal channel angular pressing (ECAP) could refine the microstructure and modify the resulting deformation texture. In this study an AZ31 strip (3Al-1Zn-Mg Bal. wt. %) produced by TRC was hot rolled at 400 °C using a rolling speed of 10 m/min. A speed ratio of 10% between rolls was used. Rolled samples with dimensions of 200 x 200 x 1.8 mm receive a single pass of ECAP at 200 °C. After processing the microstructure of the samples were analyzed by means of optical microscopy, electron backscatter diffraction and X-ray diffraction. the results showed that the microstructure can be refined and the texture altered in comparison to conventionally rolled AZ31 sheets. the samples processed by ECAP show an increase in formability up to 50% higher than conventionally rolled sheets. this demonstrates the potential of using shear deformation for processing Mg sheets. © (2015) Trans Tech Publications, Switzerland.

关键词： Equal channel angular pressing

来源：评论

学校读者我要写书评

暂无评论

Hardware Implementation of Blind Source Separation Algorithm Using ZYBO Z7& Xilinx System Generator

Hardware Implementation of Blind Source Separation Algorithm...

引用

international conference on Renewable Energies for Developing Countries (REDEC)

作者： Mohcin Mekhfioui Amal Satif Omar Mouhib Rachid Elgouri Abdelkader Hadjoudja Laamari Hlou Laboratory of Electrical Engineering and Energy Systems Faculty of Science Ibn Tofail University Kenitra Morocco

ISBN: (数字)9781728155951

ISBN: (纸本)9781728155968

Blind Source Separation (BSS) is termed as the extraction of source signals from mixed data without or with little knowledge about the source signals. It is used in many fields of research such as military, medical and industry. Among the many challenges that must be overcome to apply BSS to a real system, implementation restrictions are the most difficult. Software implementations require a significant amount of overhead to make them applicable to real-time systems and have a lower maximum speed of operation than hardware implementations. When speed is desired, the algorithm should be implemented on specialized hardware such as a Field Programmable Gate Array(FPGA) which allows the user to take advantage of parallel computational abilities. this paper describes the methodology for implementing real-time DSP applications on FPGA and the concept of hardware software cosimulation of the Second Order Blind Identification (SOBI) algorithm using MATLAB Simulink and Xilinx System Generator (XSG). Performances of efficient architectures are implemented on FPGA ZYBO Z7 Zynq-7020.

关键词： Field programmable gate arrays Blind source separation Generators Signal processing algorithms Hardware Real-time systems Software packages

来源：评论

学校读者我要写书评

暂无评论

Combining Computer Vision and Word processing to Classify Film Genres

Combining Computer Vision and Word Processing to Classify Fi...

引用

Digital Signal processing and its Applications (DSPA), international conference on

作者： Nikita Andriyanov Vitaly Dementev Data Analysis and Machine Learning Department Financial University under the Government of the Russian Federation Moscow Russia Radio Engineering Telecommunications and Information Security Department Ulyanovsk State Technical University Ulyanovsk Russia

the paper investigates various deep learning approaches for classifying movies by genre. To prepare the initial dataset, a system for parsing textual information and images from the Internet Movie Database, IMDB (https://***/) was implemented. Pandas (Python library) methods were applied to generate the dataset. Particular attention was paid to the preparation of target information and the choice of the learning loss function due to the fact that the problem was reduced to a multi-label classification. A number of transfer learning algorithms have been implemented for convolutional neural network models and transformer architectures when processing movie posters, as well as simple machine learning models and deep learning transformer-type models for text classification. the paper proposes a method for complexing extracted features from heterogeneous data in another neural network to improve classification accuracy. It was possible to achieve an increase in the average cosine distance for the test sample by 3% higher than when using standard models and about 7% in sense of intersection over union metric.

关键词： Measurement Deep learning Computer vision Computational modeling Computer architecture Motion pictures Transformers

来源：评论

学校读者我要写书评

暂无评论

A C++ Library for Memory Layout and Performance Portability of Scientific Applications 28th

A C++ Library for Memory Layout and Performance Portabilit...

引用

28th international European conference on parallel and Distributed Computing , Euro-Par 2022

作者： Incardona, Pietro Gupta, Aryaman Yaskovets, Serhii Sbalzarini, Ivo F. Technische Universität Dresden Faculty of Computer Science Dresden Germany Max Planck Institute of Molecular Cell Biology and Genetics Dresden Germany Center for Systems Biology Dresden Dresden01307 Germany Center for Scalable Data Analytics and Artificial Intelligence ScaDS.AI Dresden/Leipzig Germany

ISBN: (纸本)9783031312083

We present a C++14 library for performance portability of scientific computing codes across CPU and GPU architectures. Our library combines generic data structures like vectors, multi-dimensional arrays, maps, graphs, and sparse grids with basic, reusable algorithms like convolutions, sorting, prefix sum, reductions, and scan. the memory layout of the data structures is adapted at compile-time using tuples with optional memory mirroring between CPU and GPU. We combine this transparent memory mapping with generic algorithms under two alternative programming interfaces: a CUDA-like kernel interface for multi-core CPUs, Nvidia GPUs, and AMD GPUs, as well as a lambda interface. We validate and benchmark the presented library using micro-benchmarks, showing that the abstractions introduce negligible performance overhead, and we compare performance against the current state of the art. © 2023, the Author(s), under exclusive license to Springer Nature Switzerland AG.

关键词： Graphics processing unit

来源：评论

学校读者我要写书评

暂无评论

All You Should Know About Large Language Models (LLMs)

All You Should Know About Large Language Models (LLMs)

引用

international conference on Advanced Communication Technologies and Networking (CommNet)

作者： Abadila Alaktif Meriyem Chergui Imane Daoudi Abdelkrim Ammoumou C3S ENSEM EST Hassan II University Casablanca Morocco

ISBN: (数字)9798350367027

ISBN: (纸本)9798350367034

this article examines recent advancements in Large Language Models (LLMs), focusing on efficiency improvements and practical implementation. through analysis of literature from 2021 to present, we identify key innovations making these models more accessible. We highlight three main areas of progress: enhanced Transformer architectures enabling longer text processing with fewer resources, novel fine-tuning techniques for efficient task adaptation, and expanded applications beyond language processing into vision, speech, and reinforcement learning. the review evaluates efficiency innovations including patternbased approaches and dynamic attention mechanisms, comparing their benefits and trade-offs. Our analysis provides insights for researchers and practitioners working to make LLMs more practical and accessible, particularly addressing deployment challenges like memory and computational constraints.

关键词： Adaptation models Technological innovation Reviews Computational modeling Large language models Focusing Speech enhancement Transformers Computational efficiency Text processing

来源：评论

学校读者我要写书评

暂无评论

Hardware models for automated partitioning and mapping in multi-core systems

Hardware models for automated partitioning and mapping in mu...

引用

IEEE international Workshop on Intelligent Data Acquisition and Advanced Computing Systems: Technology and Applications

作者： Lukas Krawczyk Erik Kamsties University of Applied Sciences Fachhochschule Dortmund Germany

Multi-core CPUs offer several major benefits in mobile and embedded systems. For instance, they provide better energy efficiency and more computing power at the same clock speed compared to single-core CPUs. these benefits do not come for free: A program has to be divided into tasks, which can be executed in parallel on different cores. Partitioning of software and mapping on cores are nontrivial activities that require detailed knowledge about the multi-core CPU, e.g., if shared memories are available, their size, and other properties. Such programmers' information about a CPU is typically stored in processor handbooks. If this information would be available in a machine readable model, we call it hardware model, the partitioning and mapping activities can be automated. In this paper, we propose a hardware model and illustrate it using an example of a Freescale multi-core CPU. We then discuss a small case study, which illustrates the use of the hardware model in partitioning, mapping, and code generation.

关键词： Hardware Embedded systems Data models System-on-chip Unified modeling language Multicore processing

来源：评论

学校读者我要写书评

暂无评论

Performance Analysis of parallel FIR Digital Filter Based on Even Symmetric Fast FIR Algorithm using Different Adders

Performance Analysis of Parallel FIR Digital Filter Based on...

引用

international conference of Electronics, Communication and Aerospace Technology (ICECA)

作者： S Narendran B T Geetha Institute of Electronics and Communication Engineering Saveetha School of Engineering SIMATS Chennai India

ISBN: (纸本)9781665435253

the ultimate goal of the research work is to analyze the performance of FIR filter designed based on FFA algorithm. the proposed structure aims at area-optimized adders which are designed for symmetric coefficient that reduces hardware cost in the design with a constrain that the filter tap is of even order. Area analysis of silicon chip is performed using different adders like Ripple Carry Adder (RCA), Carry Look-Ahead Adder (CLA) and Conditional Sum Adder (COS). Performance analysis report for a 3 parallel 96-tap filter is considered based on the above adder, where the RCA requires less area utilization compared to other three adders. the symmetric convolution-based RCA Fast FIR Algorithm (FFA) structure requires only 896 slices whereas COS requires 1024 slices and CLA requires 960 slices. For higher-order, FIR filter structure of 1024 or higher taps, the savings in the area utilization for RCA based FFA is approximately 10% and 15% compared to CLA and COS basedFFA respectively because COS requires more hardware building blocks. the RCA based FFA increases the latency by approximately 9% and 7% compared to CLA and COS based FFA respectively, and there will be a trade-off.

关键词： Costs Finite impulse response filters Signal processing algorithms Digital signal processing Filtering algorithms Approximation algorithms Hardware

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：