检索结果-内蒙古大学图书馆

Marmotini: A Weight Density Adaptation Architecture With Hybrid compression Method for Spiking Neural Network

IEEE TRANSACTIONS ON VERY LARGE SCALE INTEGRATION (VLSI) SYSTEMS 2024年第12期32卷 2293-2302页

作者： Wang, Zilin Zhong, Yi Ou, Zehong Yang, Youming Feng, Shuo Chen, Guang Cui, Xiaoxin Jia, Song Wang, Yuan Peking Univ MPW Ctr Sch Integrated Circuits Key Lab Microelect Devices & Circuits Beijing 100871 Peoples R China Peking Univ Sch Software & Microelect Beijing 102600 Peoples R China Peking Univ Beijing Adv Innovat Ctr Integrated Circuits Beijing 100871 Peoples R China

Brain-inspired spiking neural network (SNN) has recently attracted widespread interest owing to its event-driven nature and relatively low-power hardware for transmitting highly sparse binary spikes. To further improve energy efficiency, some matrix compression algorithms are used for weight storage. However, the weight sparsity of different layers varies greatly. For a multicore neuromorphic system, it is difficult for the same compression algorithm to adapt to all the layers of SNN model. In this work, we propose a weight density adaptation architecture with hybrid compression method for SNN, named Marmotini. It is a multicore heterogeneous design, including three types of cores to complete computation of different weight sparsity. Benefiting from the hybrid compression method, Marmotini minimizes the waste of neurons and weights as much as possible. Besides, for better flexibility, a reconfigurable core that can be configured to compute convolutional layer or fully connected layer is proposed. Implemented on Xilinx Kintex UltraScale XCKU115 field-programmable gate array (FPGA) board, Marmotini can operate at 150-MHz frequency, achieving 244.6-GSOP/s peak performance and 54.1-GSOP/W energy efficiency at 0% spike sparsity.

关键词： Multicore processing Neurons Computer architecture compression algorithms Indexes Convolution Synapses Convolutional neural network field-programmable gate array (FPGA) neuromorphic computing spiking neural network (SNN)

来源：评论

学校读者我要写书评

暂无评论

STATISTICAL DISTRIBUTIONS OF DCT COEFFICIENTS AND THEIR APPLICATION TO AN INTERFRAME compression ALGORITHM FOR 3-D MEDICAL IMAGES

引用

IEEE TRANSACTIONS ON MEDICAL IMAGING 1993年第3期12卷 478-485页

作者： LEE, H KIM, YM ROWBERG, AH RISKIN, EA UNIV WASHINGTON DEPT RADIOLSEATTLEWA 98195

We present displacement estimated interframe (DEI) coding, a new coding scheme for 3-D medical image data sets such as x-ray CT or MR images. To take advantage of the correlation between contiguous slices, a displacement-compensated difference image based on the previous image is encoded. We have determined the best fitting distribution functions for the discrete cosine transform (DCT) coefficients obtained from displacement compensated difference images, and used the results in allocating bits and optimizing quantizers for the coefficients. The DEI scheme is compared with a 2-D block DCT as well as a full-frame DCT using the bit allocation technique by Lo and Huang [1]. For x-ray CT head images, our bit allocation and quantizer design using an appropriate distribution model resulted in a 13 dB improvement in the SNR compared to the full-frame DCT using the bit allocation technique. For an image set with 5 mm slice thickness, the DEI method compared to the 2-D block DCT gave about 5 percent improvement in the compression ratio on average and less blockiness at the same distortion. The performance gain increases to about 10 percent when the slice thickness decreases to 3 mm.

关键词： compression ratio difference image compression algorithms interframe discrete cosine transforms Three-Dimensional Images slicing labeling scheme quantizer statistical distribution

来源：评论

学校读者我要写书评

暂无评论

FILTERING RANDOM NOISE FROM DETERMINISTIC SIGNALS VIA DATA-compression

引用

IEEE TRANSACTIONS ON SIGNAL PROCESSING 1995年第11期43卷 2595-2605页

作者： NATARAJAN, BK Hewlett Packard Laboratories Palo Alto CA USA

We present a novel technique for the design of filters for random noise, leading to a class of filters called Occam filters. The essence of the technique is that when a lossy data compression algorithm is applied to a noisy signal with the allowed loss set equal to the noise strength, the loss and the noise tend to cancel rather than add. We give two illustrative applications of the technique to univariate signals. We also prove asymptotic convergence bounds on the effectiveness of Occam filters.

关键词： Filtering Data compression Noise cancellation Wiener filter Nonlinear filters compression algorithms Additive noise Convergence Power measurement Noise measurement

来源：评论

学校读者我要写书评

暂无评论

Local Decode and Update for Big Data compression

引用

IEEE TRANSACTIONS ON INFORMATION THEORY 2020年第9期66卷 5790-5805页

作者： Vatedka, Shashank Tchamkerten, Aslan Indian Inst Technol Dept Elect Engn Hyderabad 502285 India Inst Polytech Paris Dept Communicat & Elect Telecom Paris F-91120 Palaiseau France

This paper investigates data compression that simultaneously allows local decoding and local update. The main result is a universal compression scheme for memoryless sources with the following features. The rate can be made arbitrarily close to the entropy of the underlying source, contiguous fragments of the source can be recovered or updated by probing or modifying a number of codeword bits that is on average linear in the size of the fragment, and the overall encoding and decoding complexity is quasilinear in the blocklength of the source. In particular, the local decoding or update of a single message symbol can be performed by probing or modifying on average a constant number of codeword bits. This latter part improves over previous best known results for which local decodability or update efficiency grows logarithmically with blocklength.

关键词： Decoding Encoding Entropy Complexity theory Computational modeling Cloud computing DNA Data compression source coding compression algorithms big data applications

来源：评论

学校读者我要写书评

暂无评论

PERFORMANCE ANALYSIS OF REVERSIBLE IMAGE compression TECHNIQUES FOR HIGH-RESOLUTION DIGITAL TELERADIOLOGY

引用

IEEE TRANSACTIONS ON MEDICAL IMAGING 1992年第3期11卷 430-445页

作者： KUDUVALLI, GR RANGAYYAN, RM Department of Electrical and Computer Engineering University of Calgary Calgary AB Canada Dept. of Electr. & Comput. Eng. Calgary Univ. Alta. Canada

In recent years, there have been a number of studies addressing both reversible and irreversible compression of medical images ranging from 256 x 256 to 2048 x 2048 pixels in spatial resolution. There is a need to address the high-resolution end of the image categories, namely mammograms and chest X-rays, which require resolution of the order of 4096 x 4096 pixels. Further, data compression schemes for most medical applications have to be information-preserving or reversible. In this paper, the performance of a number of block-based, reversible, compression algorithms suitable for compression of very large-format images (4096 x 4096 pixels or more) is compared to that of a novel two-dimensional linear predictive coder developed by extending the multichannel version of the Burg algorithm to two dimensions. The compression schemes implemented are: Huffman coding, Lempel-Ziv coding, arithmetic coding, two-dimensional linear predictive coding (in addition to the aforementioned one), transform coding using discrete Fourier-, discrete cosine-, and discrete Walsh transforms, linear interpolative coding, and combinations thereof. We discuss the performance of these coding techniques with a few mammograms and chest radiographs digitized to sizes up to 4096 x 4096, 10 b pixels. We have achieved compression from 10 bits to 2.5-3.0 b/pixel on these images without any loss of information. The modified multichannel linear predictor out-performs the other methods while offering certain advantages in implementation.

关键词： compression algorithms information loss pixel image compression method coding techniques(computer programming) Teleradiology compression

来源：评论

学校读者我要写书评

暂无评论

Multi-distribution noise quantisation: an extreme compression scheme for transformer according to parameter distribution

引用

CONNECTION SCIENCE 2022年第1期34卷 990-1004页

作者： Yu, Zaiyang Li, Shuang Sun, Linjun Liu, Liang Wang Haining Chinese Acad Sci Inst Semicond Beijing Peoples R China Wave Grp Cognit Comp Technol Joint Lab Beijing Peoples R China Univ Chinese Acad Sci Ctr Mat Sci & Optoelect Engn Beijing 100049 Peoples R China Univ Chinese Acad Sci Sch Microelect Beijing 100049 Peoples R China Beijing Key Lab Semicond Neural Network Intellige Beijing 100083 Peoples R China Peoples Publ Secur Univ China Sch Police Adm Beijing Peoples R China

With the development of deep learning, neural networks are widely used in various fields, and the improved model performance also introduces a considerable number of parameters and computations. Model quantisation is a technique that turns floating-point computing into low-specific-point computing, which can effectively reduce model computation strength, parameter size, and memory consumption but often bring a considerable loss of accuracy. This paper mainly addresses the problem where the distribution of parameters is too concentrated during quantisation aware training (QAT). In the QAT process, we use a piecewise function to statistics the parameter distributions and simulate the effect of quantisation noise in each round of training, based on the statistical results. Experimental results show that by quantising the Transformer network, we lose less precision and significantly reduce the storage cost of the model;compared with the full precision LSTM network, our model has higher accuracy under the condition of a similar storage cost. Meanwhile, compared with other quantisation methods on language modelling task, our approach is more accurate. We validated the effectiveness of our policy on the WikiText-103 and PENN Treebank datasets. The experiments show that our method extremely compresses the storage cost and maintains high model performance.

关键词： compression algorithms natural language processing transformer vector quantisation appendices

来源：评论

学校读者我要写书评

暂无评论

Data-Aware Adaptive compression for Stream Processing

引用

IEEE TRANSACTIONS ON KNOWLEDGE AND DATA ENGINEERING 2024年第9期36卷 4531-4549页

作者： Zhang, Yu Zhang, Feng Li, Hourun Zhang, Shuhao Guo, Xiaoguang Chen, Yuxing Pan, Anqun Du, Xiaoyong Renmin Univ China Sch Informat Key Lab Data Engn & Knowledge Engn MOE Beijing 100872 Peoples R China Nanyang Technol Univ NTU Sch Comp Sci & Engn SCSE Singapore 639798 Singapore Tencent Inc Database R&D Dept Shenzhen Peoples R China

Stream processing has been in widespread use, and one of the most common application scenarios is SQL query on streams. By 2021, the global deployment of IoT endpoints reached 12.3 billion, indicating a surge in data generation. However, the escalating demands for high throughput and low latency in stream processing systems have posed significant challenges due to the increasing data volume and evolving user requirements. We present a compression-based stream processing engine, called CompressStreamDB, which enables adaptive fine-grained stream processing directly on compressed streams, to significantly enhance the performance of existing stream processing solutions. CompressStreamDB utilizes nine diverse compression methods tailored for different stream data types and integrates a cost model to automatically select the most efficient compression schemes. CompressStreamDB provides high throughput with low latency in stream SQL processing by identifying and eliminating redundant data among streams. Our evaluation demonstrates that CompressStreamDB improves average performance by 3.84x and reduces average delay by 68.0% compared to the state-of-the-art stream processing solution for uncompressed streams, along with 68.7% space savings. Besides, our edge trials show an average throughput/price ratio of 9.95x and a throughput/power ratio of 7.32x compared to the cloud design.

关键词： Encoding compression algorithms Real-time systems Low latency communication Data processing Costs Delays Data compaction and compression stream processing edge computing

来源：评论

学校读者我要写书评

暂无评论

Generalized Biwords for Bitext compression and Translation Spotting

引用

JOURNAL OF ARTIFICIAL INTELLIGENCE RESEARCH 2012年第1期43卷 389-418页

作者： Sanchez-Martinez, Felipe Carrasco, Rafael C. Martinez-Prieto, Miguel A. Adiego, Joaquin Univ Alacant Dept Llenguatges & Sistemes Informat E-03071 Alacant Spain Univ Valladolid Dept Informat E-47011 Valladolid Spain

Large bilingual parallel texts (also known as bitexts) are usually stored in a compressed form, and previous work has shown that they can be more efficiently compressed if the fact that the two texts are mutual translations is exploited. For example, a bitext can be seen as a sequence of biwords pairs of parallel words with a high probability of co-occurrence that can be used as an intermediate representation in the compression process. However, the simple biword approach described in the literature can only exploit one-to-one word alignments and cannot tackle the reordering of words. We therefore introduce a generalization of biwords which can describe multi-word expressions and reorderings. We also describe some methods for the binary compression of generalized biword sequences, and compare their performance when different schemes are applied to the extraction of the biword sequence. In addition, we show that this generalization of biwords allows for the implementation of an efficient algorithm to look on the compressed bitext for words or text segments in one of the texts and retrieve their counterpart translations in the other text-an application usually referred to as translation spotting-with only some minor modifications in the compression algorithm.

关键词： compression algorithms Generalization Word COMPRESSED texts REORDERING efficient algorithm compression Metrorrhagia

来源：评论

学校读者我要写书评

暂无评论

A Rolling Hash Algorithm and the Implementation to LZ4 Data compression

引用

IEEE ACCESS 2020年 8卷 35529-35534页

作者： Jiang, Hao Lin, Sian-Jheng Univ Sci & Technol China Sch Informat Sci & Technol CAS Key Lab Electromagnet Space Informat Hefei 230027 Peoples R China

LZ77 is a dictionary compression algorithm by replacing the repeating sequence with the addresses of the previous referenced data in the stream. To find out these repetition, the LZ77 encoder maintains a hashing table, which have to frequently calculate hash values during the encoding process. In this paper, we present a class of rolling hash functions, that can calculate multiple hash values via a carry-less multiplication instruction. Then the proposed hash function is implemented in LZ4, which is a derivative of LZ77. The simulation shows that the encoding throughput of LZ4 has 15.7% improvement in average, and the compression ratio is +/- 1% in most cases.

关键词： Encoding Data compression compression algorithms Dictionaries Indexes Image coding Image restoration Carry-less multiplication LZ4 LZ77 rolling hash SIMD

来源：评论

学校读者我要写书评

暂无评论

Vessel AIS Trajectory Online compression Based on Scan-Pick-Move Algorithm Added Sliding Window

引用

IEEE ACCESS 2020年 8卷 109350-109359页

作者： Sun, Shuang Chen, Yan Piao, Zaiji Zhang, Jinsong Dalian Maritime Univ Sch Maritime Econ & Management Dalian 116026 Peoples R China Dalian Maritime Univ Sch Marine Elect Engn Dalian 116026 Peoples R China

The trajectory data of vessel AIS (automatic identification system) has important theoretical and application value for information supporting decisions. However, large sizes lead to difficulties in storing, querying, and processing. To solve the problems of high compression ratio and longtime consumption of the existing online trajectory compression algorithm, an SPM (scan-pick-move) trajectory data compression algorithm added sliding window is proposed. In order to better compress vessel trajectory data regarding compression efficiency, the sliding window is added to the classical SPM algorithm. In order to reduce trajectory data storage space, the maximum offset distance reference trajectory point is used as the criterion of whether the current trajectory point can be compressed. In this paper, the multi-dimensional space-time characteristics of trajectory data, such as distance error, compression ratio and compression time, are selected to evaluate the trajectory compression method from three levels: geometric characteristics, motion characteristics and compression efficiency. Compared with the existing SPM trajectory data compression algorithm, parallel experiments are conducted based on AIS data gathered over the duration of a month in the Japan Osaka Bay. The SPM trajectory compression algorithm added sliding window can significantly reduce the compression time and outperforms other existing trajectory compression algorithms in term of average compression error at high compression strengths. Also, the proposed method has high compression efficiency in the range of commonly used compression thresholds.

关键词： Trajectory Artificial intelligence compression algorithms Marine vehicles Data compression Microsoft Windows Windows AIS data compression ratio scan-pick-move sliding window vessel trajectory compression

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

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

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：