检索结果-内蒙古大学图书馆

57th Asilomar Conference on Signals, Systems and Computers

作者： Culp, Michael Arizona State Univ Sch Math & Stat Sci Tempe AZ 85287 USA Naval Air Warfare Ctr Weap Div China Lake CA 93555 USA

ISBN: (纸本)9798350325744

sparse coding (SC) is powerful tool for representing data in a reduced dimensionality with minimal loss of information. sparse Dictionary Learning (SDL) is the machine learning process of improving representational features for SC. Traditional frame metrics have an underlying assumption that the distribution of the dataset's spectrum is radially symmetric. Under the assumption of a radially symmetric distribution, optimized frame metrics provide an optimal sparse representation. Real-world datasets typically aren't symmetric, to which optimized frame metrics harm the performance of sparse representations. Partitioning of a dataset into balanced spectral subspaces can approximate a breakdown of the data into more radially symmetric distributions. Partitioned Dictionary Learning (PDL) utilizes the balanced subspaces to learn incoherent dictionaries to improve sparse representations. The work of PDL serves as a basis to extend the popular Convolutional Dictionary Learning (CDL) into a Partitioned Convolutional Dictionary Learning (PCDL), where the spectral partitioning is efficient to compute.

关键词： sparse Representations sparse coding Sparsity Coherence Dictionary Learning Spectral Analysis

来源：评论

学校读者我要写书评

暂无评论

Dictionary Learning for the Almost-Linear Sparsity Regime 34

Dictionary Learning for the Almost-Linear Sparsity Regime

引用

34th International Conference on Algorithmic Learning Theory (ALT)

作者： Novikov, Alexei White, Stephen Penn State Univ Dept Math University Pk PA 16802 USA

Dictionary learning, the problem of recovering a sparsely used matrix D is an element of R-MxK and N independent Kx1 s-sparse vectors X is an element of R-KxN from samples of the formY = DX, is of increasing importance to applications in signal processing and data science. Early papers on provable dictionary learning identified that one can detect whether two samples y(i), y(j) share a common dictionary element by testing if their inner product (correlation) exceeds a certain threshold: vertical bar < y(i), y(j)>vertical bar > tau. These correlation-based methods work well when sparsity is small, but suffer from declining performance when sparsity grows faster than v M;as a result, such methods were abandoned in the search for dictionary learning algorithms when sparsity is nearly linear in M. In this paper, we revisit correlation-based dictionary learning. Instead of seeking to recover individual dictionary atoms, we employ a spectral method to recover the subspace spanned by the dictionary atoms in the support of each sample. This approach circumvents the primary challenge encountered by previous correlation methods, namely that when sharing information between two samples it is difficult to tell which dictionary element the two samples share. We prove that under a suitable random model the resulting algorithm recovers dictionaries in polynomial time for sparsity linear in M up to log factors. Our results improve on the best known methods by achieving a decaying error bound in dimension root M;the best previously known results for the overcomplete (K > M) setting achieve polynomial time in the linear regime only for constant error bounds. Numerical simulations confirm our results.

关键词： Compressed Sensing Dictionary Learning Sparsity sparse coding

来源：评论

学校读者我要写书评

暂无评论

Learning Asymmetric and Local Features in Multi-Dimensional Data Through Wavelets With Recursive Partitioning

引用

IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE 2022年第11期44卷 7674-7687页

作者： Li, Meng Ma, Li Rice Univ Dept Stat Houston TX 77025 USA Duke Univ Dept Stat Sci Durham NC 27708 USA

Effective learning of asymmetric and local features in images and other data observed on multi-dimensional grids is a challenging objective critical for a wide range of image processing applications involving biomedical and natural images. It requires methods that are sensitive to local details while fast enough to handle massive numbers of images of ever increasing sizes. We introduce a probabilistic model-based framework that achieves these objectives by incorporating adaptivity into discrete wavelet transforms (DWT) through Bayesian hierarchical modeling, thereby allowing wavelet bases to adapt to the geometric structure of the data while maintaining the high computational scalability of wavelet methods-linear in the sample size (e.g., the resolution of an image). We derive a recursive representation of the Bayesian posterior model which leads to an exact message passing algorithm to complete learning and inference. While our framework is applicable to a range of problems including multi-dimensional signal processing, compression, and structural learning, we illustrate its work and evaluate its performance in the context of image reconstruction using real images from the ImageNet database, two widely used benchmark datasets, and a dataset from retinal optical coherence tomography and compare its performance to state-of-the-art methods based on basis transforms and deep learning.

关键词： Wavelet analysis Bayes methods Computational modeling Data models Discrete wavelet transforms Analytical models Indexes Bayesian inference hierarchical models multi-resolution analysis image processing sparse coding wavelets

来源：评论

学校读者我要写书评

暂无评论

Cotton Fabric Defect Detection Based on K-SVD Dictionary Learning

引用

JOURNAL OF NATURAL FIBERS 2022年第15期19卷 10764-10779页

作者： Wu, Ying Lou, Lin Wang, Jun Zhejiang Sci Tech Univ Sch Fash Design & Engn Hangzhou Zhejiang Peoples R China Zhejiang Prov Engn Lab Clothing Digital Technol Hangzhou Zhejiang Peoples R China Donghua Univ Coll Text Shanghai Peoples R China Donghua Univ Key Lab Text Sci & Technol Minist Educ Shanghai Peoples R China

To improve the detection for different fabric types and defect kinds, an approach based on K-Singular Value Decomposition (K-SVD) dictionary learning method is designed to detect fabric defects, which remains important and challenging in the field of textile engineering. The proposed method has two main parts, namely the training and detection. The 32 defect-free fabric samples were used to train the dictionary by K-SVD. The detection process mainly consists of image segmentation, reconstruction by the trained dictionary, and detection. To improve the detection speed, the patch size was applied for detecting the defects, and the learned dictionaries were trained offline. For selecting general patch size, three patch sizes of 16 x 16 26 x 26, and 36 x 36 were used in the experiment. A comparative study found that the 26 x 26 has a comprehensive performance. The size and location of the defects were denoted by the black rectangle. Experimental results on 55 fabric samples demonstrated that the proposed method can efficiently detect different kinds of fabric defects and fabric types based on false detection rate and the correct detection rate.

关键词： Defect detection sparse coding K-SVD dictionary learning woven fabric threshold texture

来源：评论

学校读者我要写书评

暂无评论

Dictionary-Learning (DL)-Based sparse CSI Estimation in Multiuser Terahertz (THz) Hybrid MIMO Systems Under Hardware Impairments and Beam-Squint Effect

引用

IEEE ACCESS 2022年 10卷 113699-113714页

作者： Maity, Priyanka Srivastava, Suraj Khatri, Sunaina Jagannatham, Aditya K. Indian Inst Technol Kanpur Dept Elect Engn Kanpur 218016 Uttar Pradesh India Qualcomm Bengaluru 560056 India

This work conceives dictionary-learning (DL)-based sparse channel estimation schemes for multi-user Terahertz (THz) hybrid MIMO systems incorporating also non-idealities such as hardware impairments and beam-squint effect. Due to the presence of large antenna arrays coupled with frequency selectivity, beam squint effect is significant in THz systems. Moreover, the manufacturing and calibration errors that inevitably arise during the production of antenna arrays result in hardware impairments such as irregular antenna spacing, mutual coupling and antenna gain/phase errors in practical THz systems. To overcome these problems, this work proposes a DL algorithm to determine the best sparsifying dictionary from the acquired observations for a single-carrier frequency domain equalization (SC-FDE)-based wideband THz system in the presence of hardware impairments as well as the beam squint effect. The dictionary thus obtained is subsequently employed to exploit the sparsity of the MIMO THz channel toward CSI estimation. Furthermore, the Cramer-Rao lower bound (CRLB) is also derived for the joint DL and CSI estimation algorithm, which acts as a benchmark for the mean-squared error (MSE) performance of the channel estimate obtained. The scheme is also extended to SC-FDE-based wideband THz MIMO systems with multiple antenna users. Simulation results are presented to corroborate our analytical findings and also demonstrate the improved performance with respect to the agnostic scheme that ignores the non-idealities.

关键词： Channel estimation Antenna arrays Millimeter wave communication Wideband Dictionaries Estimation Learning systems MIMO communication Terahertz communications Dictionary learning (DL) Terahertz MIMO channel estimation hardware impairments array calibration sparse coding beam squint effect

来源：评论

学校读者我要写书评

暂无评论

A Novel Supervised Distributed Dictionary Learning based on Learned K-SVD for Image Denoising 16

A Novel Supervised Distributed Dictionary Learning based on ...

引用

16th IEEE International Symposium on Embedded Multicore/Many-core Systems-on-Chip (MCSoC)

作者： Zhang, Chaoran Huang, Huakun Zhao, Lingjun Xu, Chenkai Zhao, Rui Guangzhou Univ Cyber Space Inst Adv Technol Guangzhou Guangdong Peoples R China Guangzhou Univ Comp Sci & Cyber Engn Guangzhou Guangdong Peoples R China Guangdong Polytech Normal Univ Elect & Informat Guangzhou Guangdong Peoples R China

ISBN: (纸本)9798350393613

With limited computational resources and storage space on SoCs, deploying large deep-learning networks (DNNs) is challenging. However, dictionary learning (DicL) has lower complexity and storage space requirements while improving the interpretability of the model. In addition, deep unfolding techniques can construct high-performance end-to-end networks based on DicL. Therefore, in this paper, we apply Deep KSVD (LKSVD), a deep unfolding network based on the classical KSVD algorithm, with a distributed framework and propose a distributed dictionary learning (DDL) method called DDL-LKSVD. We experimentally validate DDL-LKSVD on the classical and fundamental image denoising problem, and the experimental results show that the average PSNR values we achieve in the DDL-LKSVD proposed in this paper on the Set12 dataset with the noise level 25 are 2.58 dB, 2.74 dB, 0.29 dB, and 0.03 dB higher than those of OMP, ISTA, K-SVD, and LKSVD, respectively.

关键词： SoCs image denoising sparse coding distributed learning dictionary learning K-SVD

来源：评论

学校读者我要写书评

暂无评论

Review of bio-inspired image sensors for efficient machine vision

引用

Advanced Photonics 2024年第2期6卷 7-19页

作者： Wenhao Tang Qing Yang Hang Xu Yiyu Guo Jiqiang Zhang Chunfang Ouyang Leixin Meng Xu Liu Zhejiang Laboratory Research Center for Frontier Fundamental StudiesHangzhouChina Zhejiang University College of Optical Science and EngineeringState Key Laboratory of Extreme Photonics and InstrumentationHangzhouChina ZJU-Hangzhou Global Scientific and Technological Innovation Center HangzhouChina Shanghai Jiao Tong University Chip Hub for Integrated Photonics Xplore(CHIPX)WuxiChina

With the rapid development of sensor networks,machine vision faces the problem of storing and computing massive *** human visual system has a very efficient information sense and computation ability,which has enlightening significance for solving the above problems in machine *** review aims to comprehensively summarize the latest advances in bio-inspired image sensors that can be used to improve machine-vision processing *** briefly introducing the research background,the relevant mechanisms of visual information processing in human visual systems are briefly discussed,including layerby-layer processing,sparse coding,and neural ***,the cases and performance of image sensors corresponding to various bio-inspired mechanisms are ***,the challenges and perspectives of implementing bio-inspired image sensors for efficient machine vision are discussed.

关键词： bio-inspired image sensor machine vision layer-by-layer processing sparse coding neural adaptation

来源：评论

学校读者我要写书评

暂无评论

Conditional Uncorrelation and Efficient Subset Selection in sparse Regression

引用

IEEE TRANSACTIONS ON CYBERNETICS 2022年第10期52卷 10458-10467页

作者： Wang, Jianji Zhang, Shupei Liu, Qi Du, Shaoyi Guo, Yu-Cheng Zheng, Nanning Wang, Fei-Yue Xi An Jiao Tong Univ Inst Artificial Intelligence & Robot Xian 710049 Peoples R China Xi An Jiao Tong Univ Natl Engn Lab Visual Informat Proc & Applicat Coll Artificial Intelligence Xian 710049 Peoples R China Xi An Jiao Tong Univ Inst Artificial Intelligence & Robot Coll Artificial Intelligence Xian 710049 Peoples R China Xi An Jiao Tong Univ Sch Software Engn Xian 710049 Peoples R China Xi An Jiao Tong Univ Coll Stomatol Key Lab Shaanxi Prov Craniofacial Precis Med Res Xian 710004 Peoples R China Chinese Acad Sci Inst Automat State Key Lab Management & Control Complex Syst Beijing 100190 Peoples R China Macau Univ Sci & Technol Inst Syst Engn Macau Peoples R China Qingdao Acad Intelligent Ind Qingdao 266109 Peoples R China

Given m d-dimensional responsors and n d-dimensional predictors, sparse regression finds at most k predictors for each responsor for linear approximation, 1 <= k <= d-1. The key problem in sparse regression is subset selection, which usually suffers from high computational cost. In recent years, many improved approximate methods of subset selection have been published. However, less attention has been paid to the nonapproximate method of subset selection, which is very necessary for many questions in data analysis. Here, we consider sparse regression from the view of correlation and propose the formula of conditional uncorrelation. Then, an efficient nonapproximate method of subset selection is proposed in which we do not need to calculate any coefficients in the regression equation for candidate predictors. By the proposed method, the computational complexity is reduced from O([1/6]k(3)+(m+1)k(2)+mkd) to O([1/6]k(3)+[1/2](m+1)k(2)) for each candidate subset in sparse regression. Because the dimension d is generally the number of observations or experiments and large enough, the proposed method can greatly improve the efficiency of nonapproximate subset selection. We also apply the proposed method in real scenarios of dental age assessment and sparse coding to validate the efficiency of the proposed method.

关键词： Correlation Matching pursuit algorithms Approximation algorithms Robots Multivariate regression Linear approximation Encoding Conditional uncorrelation dental age assessment multivariate correlation sparse coding sparse regression subset selection

来源：评论

学校读者我要写书评

暂无评论

Adaptive Central Frequencies Locally Competitive Algorithm for Speech

Adaptive Central Frequencies Locally Competitive Algorithm f...

引用

2025 IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2025

作者： Bahadi, Soufiyan Plourde, Eric Rouat, Jean NECOTIS Research Lab Université de Sherbrooke SherbrookeQC Canada

ISBN: (纸本)9798350368741

Neuromorphic computing, inspired by nervous systems, revolutionizes information processing with its focus on efficiency and low power consumption. Using sparse coding, this paradigm enhances processing efficiency, which is crucial for edge devices with power constraints. The Locally Competitive Algorithm (LCA), adapted for audio with Gammatone and Gammachirp filter banks, provides an efficient sparse coding method for neuromorphic speech processing. Adaptive LCA (ALCA) further refines this method by dynamically adjusting modulation parameters, thereby improving reconstruction quality and sparsity. This paper introduces an enhanced ALCA version, the ALCA Central Frequency (ALCA-CF), which dynamically adapts both modulation parameters and central frequencies, optimizing the speech representation. Evaluations show that this approach improves reconstruction quality and sparsity while significantly reducing the power consumption of speech classification, without compromising classification accuracy, particularly on Intel's Loihi 2 neuromorphic chip. © 2025 IEEE.

关键词： adaptive locally competitive algorithm central frequencies adaptation sparse coding speech classification

来源：评论

学校读者我要写书评

暂无评论

Spiking LCA in a Neural Circuit with Dictionary Learning and Synaptic Normalization 23

Spiking LCA in a Neural Circuit with Dictionary Learning and...

引用

Annual Neuro-Inspired Computational Elements Conference (NICE)

作者： Arana, Diego Chavez Renner, Alpha Sornborger, Andrew Los Alamos Natl Lab Los Alamos NM 87545 USA New Mexico State Univ Las Cruces NM 88003 USA Univ Zurich Inst Neuroinformat Zurich Switzerland

ISBN: (纸本)9781450399470

The Locally Competitive Algorithm (LCA) [17, 18] was put forward as a model of primary visual cortex [14, 17] and has been used extensively as a sparse coding algorithm for multivariate data. LCA has seen implementations on neuromorphic processors, including IBM's TrueNorth processor [10], and Intel's neuromorphic research processor, Loihi, which show that it can be very efficient with respect to the power resources it consumes [8]. When combined with dictionary learning [13], the LCA algorithm encounters synaptic instability [24], where, as a synapse's strength grows, its activity increases, further enhancing synaptic strength, leading to a runaway condition, where synapses become saturated [3, 15]. A number of approaches have been suggested to stabilize this phenomenon [1, 2, 5, 7, 12]. Previous work demonstrated that, by extending the cost function used to generate LCA updates, synaptic normalization could be achieved, eliminating synaptic runaway [7]. It was also shown that the resulting algorithm could be implemented in a firing rate model [7]. Here, we implement a probabilistic approximation to this firing rate model as a spiking LCA algorithm that includes dictionary learning and synaptic normalization. The algorithm is based on a synfire-gated synfire chain-based information control network in concert with Hebbian synapses [16, 19]. We show that this algorithm results in correct classification on numeric data taken from the MNIST dataset. LA-UR-22-33004

关键词： Hebbian learning synaptic stability synaptic saturation synfire-gated synfire chains sparse coding

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：