检索结果-内蒙古大学图书馆

symposium of image, signal processing, and Artificial Vision (STSIVA)

作者： Ai-Hua Zhou Li-Peng Zhu Meng-Han Xu Sen Pan Jun-Feng Qiao Hong-Bin Qiu Song Deng Global Energy Interconnection Research Institute Co. LTD Artificial Intelligence on Electric Power System State Grid Corporation Joint Laboratory (GEIRI) Beijing China State Grid Jiangsu Electric Power Co. LTD. Information & Telecommunication Branch Nanjing China Institute of Advanced Technology Nanjing University of Posts & Telecommunications Nanjing China

ISBN: (纸本)9781665426565

To solve the problem of non-synchronization between enterprise application development and data development, this paper puts forward the concept of data middle platform, which combines the two data processing mechanisms of online analytical processing (OLAP) and online transaction processing (OLTP), so that faster and better data services can be provided to the foreground business. On this basis, this paper summarizes the research status of the related technologies of the data middle platform, including the architecture of the data middle platform and the key technologies of constructing the data middle platform. In-depth analysis of the business scale and business characteristics of OLTP and OLAP in various application scenarios, focusing on the technical difficulties in the application process of OLTP and OLAP in the application scenario. Finally, it summarizes the challenges faced by the basic research from three aspects: the construction of data middle platform, data quality assurance, and the application of mixed-thing analytical data management.

关键词： Data integrity Conferences Focusing Computer architecture Data processing Market research Data models

来源：评论

学校读者我要写书评

暂无评论

A Deep Learning Method for Denoising Based on a Fast and Flexible Convolutional Neural Network

引用

ieee TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING 2022年 60卷 1页

作者： Li, Wenda Liu, Hong Wang, Jian Chinese Acad Sci Inst Geol & Geophys Key Lab Petr Resources Res Beijing 100029 Peoples R China Chinese Acad Sci Inst Acoust Beijing 100049 Peoples R China

Seismic data denoising has always been an indispensable step in the seismic exploration workflow. The quality of the results directly affects the results of subsequent inversion and migration imaging. In this article, we proposed a fast and flexible convolutional neural network (FFCNN) based on DnCNN. In contrast to the existing DnCNN and other artificial intelligence (AI)-based denoisers, FFCNN enjoys several desirable properties: 1) downsampling and upscaling operations, which can sensibly reduce runtimes and memory requirements while maintaining the denoising performance, and 2) we introduced noised level maps, which can make that a single convolutional neural network (CNN) model is expected to inherit the flexibility of handling noise models with different parameters, even spatially variant noises by noting M can be nonuniform. Another benefit of increasing the noise-level map is that we can preserve more useful seismic data information by controlling the tradeoff of noise removal effect and seismic data detail preservation. For real seismic data denoised work, the main work and advantages of this article are concentrated on the following two aspects: 1) we introduced a data augmentation strategy to overcome the lack of well-labeled samples and 2) transfer learning has been introduced to the training processing, which used the well-trained synthetic seismic data denoising network as a pretrained model. In this way, we can greatly accelerate and optimize the learning efficiency of the training network. Ultimately, we can greatly improve the computational efficiency and denoising performance based on this intelligent denoised network FFCNN. Finally, numerical experiments prove the effectiveness of our method in synthetic and real seismic data.

关键词： Convolutional neural networks Noise reduction Neural networks Training signal to noise ratio Convolution image denoising Geophysical data geophysics computing

来源：评论

学校读者我要写书评

暂无评论

A Light-Weight image to Description Model Using Knowledge Distillation

A Light-Weight Image to Description Model Using Knowledge Di...

引用

Artificial intelligence, Blockchain, and Internet of Things (AIBThings), ieee International Conference on

作者： Eisha Rehan Shahwaiz Memon Muhsin Raza M. Asif Naeem Fazeel Nadeem Bhatti Department of Computer Science National University of Computer and Emerging Sciences Islamabad Pakistan Upstart Commerce Islamabad Pakistan

ISBN: (数字)9798331529529

ISBN: (纸本)9798331529536

Generative models have revolutionized automation by significantly reducing the manual effort required for various tasks. Despite their success, the increasing size of these generative AI models poses challenges for their commercial deployment. The paper proposes a novel lightweight model, EcoLLM-Mini, by compressing large model, LLaMA2, using the technique of knowledge distillation with minimal impact on performance. Our experiments demonstrate that distilling LLaMA2 with 7B parameters into a student model with only 247 M parameters yields comparable results with almost $\mathbf{96\%}$ reduction in computation cost, $\mathbf{93.09\%}$ reduction in memory usage, and $\mathbf{88\%}$ reduction in inference time. To showcase the practical applications of our approach, we implemented an image-to-description model with two components: caption generation and description generation. Initially, our approach employs BLIP-2 to generate captions for product images. These captions are then used as input to EcoLLM-Mini, which generates meaningful descriptions of the products. This model is designed to streamline the merchandising process in the eCommerce sector.

关键词： Costs Quantization (signal) Biological system modeling computational modeling Scalability Refining Streaming media Natural language processing Electronic commerce Optimization

来源：评论

学校读者我要写书评

暂无评论

Fast encoding method for vector quantization by dynamically constructing subvectors

Fast encoding method for vector quantization by dynamically ...

引用

Midwest symposium on Circuits and Systems (MWSCAS)

作者： Zhibin Pan K. Kotani T. Ohmi New Industry Creation Hatchery Center University of Tohoku Sendai Japan Graduate School of Engineering Department of Electronic Engineering University of Tohoku Sendai Japan

The encoding speed of vector quantization (VQ) is an important problem for VQ's practical applications. Because a k-dimensional (k-D) vector can also be mathematically viewed as a k-element set so that the statistical analysis methods can be directly applied to k-D vectors. In order to approximately measure the difference between two k-D vectors, by using the well-known statistical features of the sum and the variance of a k-D vector first, the IEENNS method (S. Baek, et al., ieee signal processing Letters, vol.4, pp.325-327, 1997) has been proposed to reject most of unlikely codewords for a certain input vector. Then, by dividing a k-D vector in half to generate its two corresponding (k/2)-D subvectors and then apply the IEENNS method again to each of the subvectors, a complete-version SIEENNS method (J.S. Pan, et al., ieee Trans. image processing, voL12. pp.265-270, 2003) has been proposed as well. Because the SIEENNS method still has a large memory and computational redundancy, a simplified-version enhanced ESIEENNS method (Z. Pan et al., 2005 International symposium on Circuits and Systems, pp.6332-6335, 2005) is reported recently. However, all of these subvector-based previous works just fixedly constructed its two subvectors for simplicity, which cannot guarantee a very high search performance. Instead, this paper proposes to dynamically construct the two subvectors more efficiently based on a criterion of |S/sub y,j/ - S/sub y,j/| /spl rArr/ max by offline analyzing the property of a codeword y/sub i/. Experimental results confirmed that the proposed DESIEENNS method can improve the total search efficiency to 79.9% /spl sim/ 88.7% compared to the latest ESIEENNS method for various input images.

关键词： Encoding Vector quantization image coding Euclidean distance Search methods Nearest neighbor searches Electronics industry Industrial electronics Electronic mail Statistical analysis

来源：评论

学校读者我要写书评

暂无评论

A New Algorithm to Exploit Superword Level Parallelism

A New Algorithm to Exploit Superword Level Parallelism

引用

ieee International symposium on Dependable, Autonomic and Secure Computing (DASC)

作者： Peng Liu Rong-cai Zhao Wei Gao Shuai Wei State Key Laboratory of Mathematical Engineering and Advanced Computing Zhengzhou China National Digital Switching System Engineering & Technology Research Center Zhengzhou China

Nowadays SIMD extensions became ubiquitous for computational platform, being widely used in signal processing, multimedia, and scientific applications. However, compilers still have difficult in vectorizing some applications, such as computations on non-contiguous and especially interleaved data, and loop nest for which inner loops that carry all the dependencies and external loops with strided memory accesses. In order to solve these problems, we adopt Isomorphic Statements Grouping (ISG) after extending group following UD and DU chains, which group isomorphic computation statements that perform the same operation. And correspondingly, we generate vector permutation instructions according to the target ISA SIMD instructions at the SIMD code generation phase. This approach has been implemented in a compiler based on Open64 which translate C source program to C source program with SIMD intrinsic functions. Experiments with this compiler show that ISG Super Level Parallelism(ISG-SLP) can be used to compile a number of real-world kernels, achieving up to 3.11 speedup.

关键词： Vectors Arrays Finite impulse response filters Program processors Indexes Algorithm design and analysis image processing

来源：评论

学校读者我要写书评

暂无评论

High Performance Non-uniform FFT on Modern X86-based Multi-core Systems

High Performance Non-uniform FFT on Modern X86-based Multi-c...

引用

International symposium on Parallel and Distributed processing (IPDPS)

作者： Dhiraj D. Kalamkar Joshua D. Trzaskoz Srinivas Sridharan Mikhail Smelyanskiy Daehyun Kim Armando Manduca Yunhong Shu Matt A. Bernstein Bharat Kaul Pradeep Dubey Parallel Computing Lab Intel Labs Bangalore KA India Mayo Clinic Minnesota Rochester MN US Parallel Computing Laboratory INTEL Research Laboratory Santa Clara CA USA Department of Physiology and Biomedical Engineering Mayo Clinic Rochester MN USA Parallel Computing Lab Intel Labs Santa Clara CA USA Department of Radiology Mayo Clinic Rochester MN USA

The Non-Uniform Fast Fourier Transform (NUFFT) is a generalization of FFT to non-equidistant samples. It has many applications which vary from medical imaging to radio astronomy to the numerical solution of partial differential equations. Despite recent advances in speeding up NUFFT on various platforms, its practical applications are still limited, due to its high computational cost, which is significantly dominated by the convolution of a signal between a non-uniform and uniform grids. The computational cost of the NUFFT is particularly detrimental in cases which require fast reconstruction times, such as iterative 3D non-Cartesian MRI reconstruction. We propose novel and highly scalable parallel algorithm for performing NUFFT on x86-based multi-core CPUs. The high performance of our algorithm relies on good SIMD utilization and high parallel efficiency. On convolution, we demonstrate on average 90% SIMD efficiency using SSE, as well up to linear scalability using a quad-socket 40-core Intel(R) Xeon(R) E7-4870 Processors based system. As a result, on dual socket Intel(R) Xeon(R) X5670 based server, our NUFFT implementation is more than 4x faster compared to the best available NUFFT3D implementation, when run on the same hardware. On Intel(R) Xeon(R) E5-2670 processor based server, our NUFFT implementation is 1.5X faster than any published NUFFT implementation today. Such speed improvement opens new usages for NUFFT. For example, iterative multi channel reconstruction of a 240×240×240 image could execute in just over 3 minutes, which is on the same order as contemporary non-iterative (and thus less-accurate) 3D NUFFT-based MRI reconstructions.

关键词： Convolution Kernel Interpolation Magnetic resonance imaging image reconstruction Hardware Transforms

来源：评论

学校读者我要写书评

暂无评论

Where-what network 1: “Where” and “what” assist each other through top-down connections

Where-what network 1: “Where” and “what” assist each oth...

引用

ieee International Conference on Development and Learning, ICDL

作者： Zhengping Ji Juyang Weng Danil Prokhorov Embodied Intelligence Laboratory Michigan State University Lansing USA Toyota Technical Center Ann Arbor USA

This paper describes the design of a single learning network that integrates both object location (ldquowhererdquo) and object type (ldquowhatrdquo), from images of learned objects in natural complex backgrounds. The in-place learning algorithm is used to develop the internal representation (including synaptic bottom-up and top-down weights of every neuron) in the network, such that every neuron is responsible for the learning of its own signal processing characteristics within its connected network environment, through interactions with other neurons in the same layer. In contrast with the previous fully connected MILN [13], the cells in each layer are locally connected in the network. Local analysis is achieved through multi-scale receptive fields, with increasing sizes of perception from earlier to later layers. The results of the experiments showed how one type of information (ldquowhererdquo or ldquowhatrdquo) assists the network to suppress irrelevant information from background (from ldquowhererdquo) or irrelevant object information (from ldquowhatrdquo), so as to give the required missing information (ldquowhererdquo or ldquowhatrdquo) in the motor output.

关键词： Neurons Visualization Training Feature extraction image recognition Object recognition computational modeling

来源：评论

学校读者我要写书评

暂无评论

GPU-based Hybrid Parallel Logic Simulation for Scan Patterns

GPU-based Hybrid Parallel Logic Simulation for Scan Patterns

引用

International Test Conference in Asia (ITC-Asia)

作者： Liyang Lai Qiting Zhang Hans Tsai Wu-Tung Cheng Guangdong Provincial Key Laboratory of Digital Signal and Image Processing Key Laboratory of Intelligent Manufacturing Technology Shantou University Shantou Guangdong China Mentor a Siemens Business Wilsonville OR US

ISBN: (数字)9781728189444

ISBN: (纸本)9781728189451

GPGPU, general-purpose computing on graphics processing units, has been witnessed growing from a niche to a mainstream computing paradigm in the last decade. It is widely deployed in machine learning, artificial intelligence, and many scientific applications. In this article, we study gate-level logic simulation for scan test patterns, one of the key algorithmic components for test generation, fault grading and design rule check. We are exploring if GPGPU can deliver scalable performance speedup as promised by its massive parallelism. We discuss the limitations of the state of art work. With GPUs' architectural features in mind, a novel algorithm of hybrid race-tolerant parallel logic simulation is proposed to unleash its immense power of parallelization. Its unique integration of oblivious simulation and event-driven simulation leads to a scalable realization of parallel logic simulation for scan patterns. For the first time in the literature, it is demonstrated that a modest GPU can handle an industrial design of over twenty million gates with excellent performance scalability.

关键词： Logic gates Integrated circuit modeling computational modeling Graphics processing units Data models Instruction sets Circuit faults

来源：评论

学校读者我要写书评

暂无评论

Low-Bitwidth Floating Point Quantization for Efficient High-Quality Diffusion Models

Low-Bitwidth Floating Point Quantization for Efficient High-...

引用

ieee International Workshop/symposium on Workload Characterization

作者： Cheng Chen Christina Giannoula Andreas Moshovos Electrical and Computer Engineering University of Toronto Toronto Canada Computer Science University of Toronto & Vector Institute Toronto Canada Electrical and Computer Engineering University of Toronto & Vector Institute Toronto Canada

ISBN: (数字)9798350356038

ISBN: (纸本)9798350356045

Diffusion models are emerging models that generate images by iteratively denoising random Gaussian noise using deep neural networks. These models typically exhibit high computational and memory demands, necessitating effective post-training quantization for high-performance inference. Recent works propose low-bitwidth (e.g., 8-bit or 4-bit) quantization for diffusion models, however 4-bit integer quantization typically results in low-quality images. We observe that on several widely used hardware platforms, there is little or no difference in compute capability between floating-point and integer arithmetic operations of the same bitwidth (e.g., 8-bit or 4-bit). Therefore, we propose an effective floating-point quantization method for diffusion models that provides better image quality compared to integer quantization methods. We employ a floating-point quantization method that was effective for other processing tasks, specifically computer vision and natural language tasks, and tailor it for diffusion models by integrating weight rounding learning during the mapping of the full-precision values to the quantized values in the quantization process. We comprehensively study integer and floating-point quantization methods in state-of-the-art diffusion models. Our floating-point quantization method not only generates higher-quality images than that of integer quantization methods, but also shows no noticeable degradation compared to full-precision models (32-bit floating-point), when both weights and activations are quantized to 8-bit floating-point values, while has minimal degradation with 4-bit weights and 8-bit activations. Additionally, we introduce a better methodology to evaluate quantization effects, highlighting shortcomings with existing output quality metrics and experimental methodologies. Finally, as an additional potential benefit, our floating-point quantization method increases model sparsity by an order of magnitude, enabling further optimization opportu

关键词： Degradation Measurement image quality Quantization (signal) computational modeling Noise reduction Natural languages Text to image Diffusion models Optimization

来源：评论

学校读者我要写书评

暂无评论

Diff-FFT: Synthetic Lumbar Magnetic images From Frequency Domain-Based

Diff-FFT: Synthetic Lumbar Magnetic Images From Frequency Do...

引用

International Conference on computational intelligence and Communication Networks (CICN)

作者： Jie Cao Qingxuan Jiang Boshuo Li School of Computer Science Northeast Electric Power University Jilin China

ISBN: (数字)9798331505264

ISBN: (纸本)9798331505271

Magnetic resonance imaging is playing an increasingly important role in the diagnosis of lumbar disc disease. It can provide high-resolution soft tissue contrast, helping doctors accurately identify disc lesions, inflammation and changes in other related structures. However, the use of MRI is limited due to its high cost and significant operating and processing times. More importantly, MRI examinations are not allowed for people with pacemakers, other metal objects in the body such as cochlear implants and metal teeth, or people in critical condition or claustrophobic, as it may cause injury. For this reason, this paper proposes a frequency domain-based method for synthesizing lumbar magnetic images. First, we introduce a multi-scale frequency channel attention module to model global features, which helps the model more effectively process wide-area information and improve its accuracy and detail performance in generating lumbar magnetic resonance images. Then, a dual-resolution frequency domain-based FFN module is proposed. This module selectively retains important frequency information through a learnable quantization matrix W. The inverse method of JPEG compression. Experimental results show that this method is accurate and can construct MR images that are very close to the reference MR images, and it also outperforms the other three state-of-the-art methods.

关键词： Quantization (signal) image coding Inverse problems Magnetic resonance imaging Frequency-domain analysis Spine Magnetic resonance Transform coding Metals Magnetic domains

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：