检索结果-内蒙古大学图书馆

Efficient Multi-Class Probabilistic SVMs on GPUs

IEEE TRANSACTIONS ON KNOWLEDGE AND DATA ENGINEERING 2019年第9期31卷 1693-1706页

作者： Wen, Zeyi Shi, Jiashuai He, Bingsheng Chen, Jian Chen, Yawen Natl Univ Singapore Singapore 119077 Singapore Natl Univ Singapore Sch Comp Singapore 119077 Singapore South China Univ Technol Sch Software Engn Guangzhou 510630 Guangdong Peoples R China

Recently, many researchers have been working on improving other traditional machine learning algorithms (besides deep learning) using high-performance hardware such as Graphics processing Units (GPUs). The recent success of machine learning is not only due to more effective algorithms, but also more efficient systems and implementations. In this paper, we propose a novel and efficient solution to multi-class SVMs with probabilistic output (MP-SVMs) accelerated by GPUs. MP-SVMs are an important technique for many pattern recognition applications. However, MP-SVMs are very time-consuming to use, because using an MP-SVM classifier requires training many binary SVMs and performing probability estimation by combining results of all the binary SVMs. GPUs have much higher computation capability than CPUs and are potentially excellent hardware to accelerate MP-SVMs. Still, two key challenges for efficient GPU accelerations for MP-SVM are: (i) many kernel values are repeatedly computed as a binary SVM classifier is trained iteratively, resulting in repeated accesses to the high latency GPU memory;(ii) performing training or estimating probability in a highly parallel way requires a much larger memory footprint than the GPU memory. To overcome the challenges, we propose a solution called GMP-SVM which exploits two-level (i.e., binary SVM level and MP-SVM level) optimization for training MP-SVMs and high parallelism for estimating probability. GMP-SVM reduces high latency memory accesses and memory consumption through batch processing, kernel value reusing and sharing, and support vector sharing. Experimental results show that GMP-SVM outperforms the GPU baseline by two to five times, and LibSVM with OpenMP by an order of magnitude. Also, GMP-SVM produces the same SVM classifier as LibSVM.

关键词： Machine learning multi-class probabilistic SVMs graphics processing units

来源：评论

学校读者我要写书评

暂无评论

Real-Time image processing and Deep Learning 2019

Real-Time Image Processing and Deep Learning 2019

引用

Real-Time image processing and Deep Learning 2019

ISBN: (纸本)9781510626577

The proceedings contain 27 papers. The topics discussed include: fast multi-modal reuse: co-occurrence pre-trained deep learning models;deep learning for fast super-resolution reconstruction from multiple images;an efficient algorithm for fast block matching motion estimation using an adaptive threshold scheme;low exposure image frame generation algorithms for feature extraction and classification;parallel image and video self-recovery scheme with high recovery capability;learning optimal actions with imperfect images;CNN classification based on global and local features;kalman-based motion estimation in video surveillance systems for safety applications;and recent advances in integrated photonic-electronic technologies for high-speed processing and communication circuits for light-based transducers.

关键词：

来源：评论

学校读者我要写书评

暂无评论

A Heterogeneous parallel Processor for High-Speed Vision Chip

引用

IEEE TRANSACTIONS ON CIRCUITS AND systems FOR VIDEO TECHNOLOGY 2018年第3期28卷 746-758页

作者： Yang, Jie Yang, Yongxing Chen, Zhe Liu, Liyuan Liu, Jian Wu, Nanjian Chinese Acad Sci Inst Semicond Beijing 100083 Peoples R China

This paper proposes a heterogeneous parallel processor for high-speed vision chip. It contains four levels of processors with different parallelisms and complexities: processing element (PE) array processor, patch processing unit (PPU) array processor, self-organizing map (SOM) neural network processor, and dual-core microprocessor unit (MPU). The fine-grained PE array processor, middle-grained PPU array processor, and SOM neural network processor carry out image processing in pixel-parallel, patch-parallel, and distributed-parallel fashions, respectively. The MPU controls the overall system and executes some serial algorithms. The processor can improve the total system performance from low-level to high-level image processing significantly. A prototype is implemented with 64x64 PE array, 8 x 8 PPU array, 16 x 24 SOM network, and a dual-core MPU. The proposed heterogeneous parallel processor introduces a new degree of parallelism, namely, patch parallel, which is for parallel local-feature extraction and feature detection. It can flexibly perform the state-of-the-art computer vision as well as various image processing algorithms at high speed. Various complicated applications, including feature extraction, face detection, and high-speed tracking, are demonstrated.

关键词： Computer vision face detection heterogeneous high speed object tracking parallel processing single-instruction-multiple-data (SIMD) vision chip

来源：评论

学校读者我要写书评

暂无评论

Towards real-time and real-life image classification and detection using CNN: a review of practical applications requirements, algorithms, hardware and current trends

Towards real-time and real-life image classification and det...

引用

International Symposium for Design and Technology of Electronics Packages (SiiTME)

作者： Mariana Eugenia Ilas Constantin Ilas “Politehnica” University of Bucharest Telecom. and IT Bucharest Romania “Politehnica” University of Bucharest Bucharest Romania

ISBN: (数字)9781728175065

ISBN: (纸本)9781728175072

CNN are already used in production applications in some areas, whereas for many real-life applications this is not possible yet, mainly because of speed and/ or power consumption limitations. In parallel, there are sustained efforts for simplifying CNN structure and for designing new processor architectures. In this paper we analyze the requirements of inference time and power consumption for a large variety of applications, running on all type of platforms, from edge devices to servers. We also review all hardware systems which can be used for CNN implementation, GPUs, TPUs, FPGA, prototype ASICs. Finally, we correlate the performance of these devices with the real-time and power requirements and the complexity of most popular CNN architectures. Thus, we determine what types of practical applications are currently possible, quantify the gaps for the others, and discuss how these gaps can be reduced.

关键词： Power demand Real-time systems Hardware Complexity theory Automobiles Graphics processing units Computer architecture

来源：评论

学校读者我要写书评

暂无评论

Methodology for the implementation of real-time image processing systems using FPGAs and GPUs and their integration in EPICS using Nominal Device Support

引用

FUSION ENGINEERING AND DESIGN 2018年 130卷 26-31页

作者： Esquembri, S. Nieto, J. Ruiz, M. de Gracia, A. de Arcas, G. Univ Politecn Madrid Instrumentat & Appl Acoust Res Grp Madrid Spain

There is an increasing interest to improve the processing capabilities for imaging diagnostics in fusion devices. Furthermore, the continuous upgrades of the hardware systems used in these tasks require a flexible platform that obtains the maximum performance of these technologies: cameras, frame-grabbers, and parallel processing architectures using FPGAs, GPUs, and multicore CPUs. This work presents a methodology for the development of real-time image processing applications using the ITER CODAC Core System (CCS) software tools and the hardware configuration defined in the ITER fast controllers hardware catalog, by means of Camera Link cameras, Camera Link FPGA-based frame-grabbers, and NVIDIA GPUs. The integration with EPICS is provided using the Nominal Device Support approach and the IRIO tools integrated into CCS. This device support provides complete control of the camera configuration parameters, distribution of the image processing functions between the FPGA and the GPU, and efficient movement of the data between the different elements of the architecture. To validate the full development cycle, an EPICS application has been developed implementing different algorithms frequently used in fusion applications such as histogram, detection of levels, and classification of zones within an image. Results of the performance obtained are presented highlighting the differences among various configurations running the selected algorithms.

关键词： Graphics processing units (GPU) image acquisition and processing Fusion experiments Experimental Physics and Industrial Control System (EPICS) Nominal Device Support (NDS) Field Programmable Gate Array (FPGA)

来源：评论

学校读者我要写书评

暂无评论

Memory-efficient 3D connected component labeling with parallel computing

引用

SIGNAL image AND VIDEO processing 2018年第3期12卷 429-436页

作者： Ohira, Norihiro Tokyo Metropolitan Ind Technol Res Inst Sumida Ku 12th FloorKFC Bldg 1-6-1 Tokyo 1300015 Japan

Connected component labeling is a frequently used image processing task in many applications. Moreover, in recent years, the use of 3D image data has become widespread, for instance, in 3D x-ray computed tomography and magnetic resonance imaging. However, because ordinary labeling algorithms use a large amount of memory and 3D images are generally large, labeling 3D image data can cause memory shortages. Furthermore, labeling a large image is time-consuming. In this paper, we proposed new memory-efficient connected component labeling algorithm for 3D images with parallel computing. In this method, we accelerate the labeling process using parallel computing. In addition, we use a spans matrix and compressed label matrix to reduce memory usage. We also use an equivalence chain approach to speed up the calculation. Furthermore, the algorithm has two options for further processing performance or further memory savings. In the experiments on real examples, the proposed algorithm with the option for processing speed was faster and used less memory than the conventional label equivalence method. In contrast, with the proposed method using the memory-efficient option, we could further reduce memory from one-eighth to one-thirteenth that used by the label equivalence method while maintaining the same performance.

关键词： image processing Labeling Memory-efficient parallel computing

来源：评论

学校读者我要写书评

暂无评论

RADIATION-INDUCED ACOUSTIC imaging SYSTEM DEVELOPMENT AND applications

RADIATION-INDUCED ACOUSTIC IMAGING SYSTEM DEVELOPMENT AND AP...

引用

作者： SIQI WANG UNIVERSITY OF OKLAHOMA

学位级别：硕士

This thesis reports the development of three different radiation induced acoustic imaging systems excited by a pulsed laser, x-ray, and electric field respectively. Firstly, a laser induced photoacoustic imaging system for non-destructive testing (pNDT) is presented in Chapter I. The pNDT is comprised of three major components: picosecond pulsed laser based ultrasonic actuator, ultrasound receiver, data processing and computing subsystem. A CFRP composted was scanned and both the micro-scale and macro-scale damage are detected and localized with spatial resolution of 100 µm. Secondly, an x-ray-induced acoustic tomography (xACT) system was developed for bone imaging in Chapter ii. A 500 KHz ultrasound transducer was used to obtain a 3D image of a chicken bone placed underneath the -ray source. The first xACT biological bone sample was linearly scanned with 12-posiiton increments. The image was then obtained with a back-projection reconstruction algorithm. The chicken bone was successfully reconstructed, demonstrating the potentials of xACT systems in bone density in vivo imaging. Thirdly, a new imaging modality, electroacoustic tomography (EAT), was designed for electrical therapy monitoring. The acoustic signals generated by high-voltage pulsing electrical fields during electrical therapies were detected and analyzed. The optimal parameters for effective EA signal generation were found. The results demonstrated the potentials of the pNDT system in ultra-high resolution CFRP NDT, the xACT system in fast 3D bone imaging, and the EAT system in real-time electric therapy monitoring.

关键词： Engineering Electronics and Electrical. medical imaging photoacoustic imaging x-ray-induced acoustic imaging

来源：评论

学校读者我要写书评

暂无评论

Blazed grating theory to minimize the non-idealities in LCoS devices 13

Blazed grating theory to minimize the non-idealities in LCoS...

引用

Conference on Optics and Photonics for Information processing xiiI

作者： Marquez, A. Martinez-Guardiola, F. J. Frances, J. Ramirez, M. G. Calzado, E. M. Morales-Vidal, M. Gallego, S. Pascual, I. Univ Alicante Dept Fis Ingn Sistemas & T Senal POB 99 E-03080 Alicante Spain Univ Alicante IU Fis Aplicada Ciencias & Tecnol POB 99 E-03080 Alicante Spain Univ Alicante Dept Opt Farmacol & Anat POB 99 E-03080 Alicante Spain

ISBN: (纸本)9781510629660

parallel-aligned liquid crystal on silicon (PA-LCoS) microdisplays are widely used in spatial light modulation applications, especially in those requiring phase-only modulation. One such application area is programmable diffractive optics which plays a very important role in modern optical imaging systems or in optical interconnections for optical telecommunications. Among the multilevel diffractive optical elements (DOEs) we focus on the important case of the blazed gratings. We develop the corresponding analytical expressions for the diffracted field where, as one of the novelties in the work, an analytical expression including the fill factor and the flicker is obtained. This enables to have a model against to compare the experimental results in a number of situations where fill factor, flicker, period, and number of quantization levels are the variables. This also enables to design appropriate compensation techniques to enhance the performance of the blazed gratings.

关键词： Liquid crystal on silicon displays parallel aligned phase-only modulation spatial light modulation flicker fringing-field blazed gratings diffractive optics Stokes polarimetry

来源：评论

学校读者我要写书评

暂无评论

Adorym: A multi-platform generic x-ray image reconstruction framework based on automatic differentiation

arXiv

引用

arxiv 2020年

作者： Du, Ming Kandel, Saugat Deng, Junjing Huang, xiaojing Demortiere, Arnaud Nguyen, Tuan Tu Tucoulou, Remi de Andrade, Vincent Jin, Qiaoling Jacobsen, Chris Advanced Photon Source Argonne National Laboratory LemontIL60439 United States Applied Physics Program Northwestern University EvanstonIL60208 United States National Synchrotron Light Source II Brookhaven National Laboratory UptonNY11973 United States CNRS UMR 7314 Université de Picardie Jules Verne Hub de l’Energie Amiens Cedex80039 France European Synchrotron Radiation Facility Grenoble38000 France Department of Physics & Astronomy Chemistry of Life Processes Institute Northwestern University EvanstonIL60208 United States

We describe and demonstrate an optimization-based x-ray image reconstruction framework called Adorym. Our framework provides a generic forward model, allowing one code framework to be used for a wide range of imaging methods ranging from near-field holography to and fly-scan ptychographic tomography. By using automatic differentiation for optimization, Adorym has the flexibility to refine experimental parameters including probe positions, multiple hologram alignment, and object tilts. It is written with strong support for parallel processing, allowing large datasets to be processed on high-performance computing systems. We demonstrate its use on several experimental datasets to show improved image quality through parameter *** Codes 78-04 © 2020, CC BY.

关键词： image reconstruction

来源：评论

学校读者我要写书评

暂无评论

Noise removal of the x-ray medical image using fast spatial filters and GPU 41

Noise removal of the X-ray medical image using fast spatial ...

引用

Conference on applications of Digital image processing xLI

作者： Cadena, Luis Zotin, Alexander Cadena, Franklin Espinosa, Nikolai Univ Fuerzas Armadas ESPE Ave Gral Ruminahui S-N Sangolqui Ecuador Reshetnev Siberian State Univ Sci & Technol 31 Krasnoyarsky Rabochy Pr Krasnoyarsk 660037 Russia Coll Juan Suarez Chacon Quito Ecuador

ISBN: (数字)9781510620766

ISBN: (纸本)9781510620766

Medical images are corrupted by different types of noises caused by the equipment itself. It is very important to obtain precise images to facilitate accurate observations for the given application. Removing of noise from images is now a very challenging issue in the field of medical image processing. This work undertake the study of noise removal techniques in medical image by using fast implementation of different digital filters, such as average, median and Gaussian filter. processing of x-ray medical images takes a significant time. Now days modern hardware allows to use parallel technology for image processing on CPU and GPU. Using GPU processing technology were proposed parallel implementations of noise reduction algorithm taking into account the data parallelism. The experimental study conducted on medical x-ray image, so that to choose the best filters considering medical task and time of processing. The comparison of the implementation of fast filters algorithm and GPU implementation show great increase in performance. Graphics processing units (GPUs) are used today in a wide range of applications, mainly because they can dramatically accelerate parallel computing. In the field of medical imaging, GPUs are in some cases crucial for enabling practical use of computationally demanding algorithms.

关键词： medical image average fast filter median fast filter Gaussian fast filter parallel programming GPU

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：