检索结果-内蒙古大学图书馆

11th International image processing and Communications conference (IP and C)

作者： Zhao, Ting Lazendic, Srdan Zhao, Yuxin Montereale-Gavazzi, Giacomo Pizurica, Aleksandra Harbin Engn Univ Coll Automat Harbin Peoples R China Univ Ghent TELIN GAIM Dept Telecommun & Informat Proc Ghent Belgium Univ Ghent Dept Math Anal Ghent Belgium Royal Belgian Inst Nat Sci Operat Directorate Nat Environm Brussels Belgium Univ Ghent Renard Ctr Marine Geol Dept Geol Ghent Belgium

ISBN: (纸本)9783030312541;9783030312534

In this paper we develop a novel classification method for multibeam sonar images based on the Weyl transform. The texture descriptor based on Weyl coefficients describes effectively the multiscale correlation features appearing in the sonar images. Our classification approach combines the Weyl coefficients with statistical features that are commonly used in the analysis of seabed sonar images and captures the morphological variation and geoacoustic characteristics of the seafloor. We employ a neural network as a classifier. The proposed combined feature extraction method demonstrates better performance than the commonly used statistical methods in this application.

关键词： Multibeam data processing Multibeam sonar image Feature extraction Weyl transform Acoustic sediment classification

来源：评论

学校读者我要写书评

暂无评论

Dual Forward-Backward Unfolded Network for Flexible Plug-and-Play

Dual Forward-Backward Unfolded Network for Flexible Plug-and...

引用

European Signal processing conference (EUSIPCO)

作者： Audrey Repetti Matthieu Terris Yves Wiaux Jean-Christophe Pesquet Department of Actuarial Mathematics & Statistics Heriot-Watt University Edinburgh UK Signals and Systems Institute of Sensors Heriot-Watt University Edinburgh UK Université Paris-Saclay CentralcSupélec Inria Center for Visual Computing Gif sur Yvette France

ISBN: (数字)9789082797091

ISBN: (纸本)9781665467995

Proximal methods have been extensively used to find maximum a posteriori (MAP) estimates of unknown images from degraded measurement. Recently, they have been mixed with neural networks (NN) to further improve the reconstruction quality. Two approaches can be distinguished: unfolded NNs, implementing a given iteration number of an optimisation algorithm, and plug-and-play (PnP) algorithms, incorporating NNs in existing optimisation algorithms. Unfolded NNs usually incorporate the measurement operator in the learning process, which can be prohibitive for applications with non-fixed measurement operators. PnP do not have this drawback, but involved NNs still depend on the underlying statistical models (e.g., higher noise level on the measurements requires stronger denoisers). In this work, we propose a PnP algorithm based on forward-backward (FB) iterations, where the learned denoiser is an unfolded NN based on dual-FB iterations. This NN is built to mimic a Gaussian denoiser from a MAP viewpoint. This allows us to introduce a regularisation parameter in the model to tune the regularization strength, similarly to standard variational approaches. This has the advantage of making the learned NN more adaptive to a variety of inverse problem statistical models, without requiring to train the NN for different noise levels.

关键词： Adaptation models Inverse problems Signal processing algorithms Artificial neural networks Signal processing image restoration Noise measurement

来源：评论

学校读者我要写书评

暂无评论

Normalized Stability: A Cross-Level Design Metric for Early Termination in stochastic Computing

Normalized Stability: A Cross-Level Design Metric for Early ...

引用

Asia and South Pacific Design Automation conference

作者： Di Wu Ruokai Yin Joshua San Miguel University of Wisconsin-Madison Madison WI USA

ISBN: (数字)9781450379991

ISBN: (纸本)9781728180571

stochastic computing is a statistical computing scheme that represents data as serial bit streams to greatly reduce hardware complexity. The key trade-off is that processing more bits in the streams yields higher computation accuracy at the cost of more latency and energy consumption. To maximize efficiency, it is desirable to account for the error tolerance of applications and terminate stochastic computations early when the result is acceptably accurate. Currently, the stochastic computing community lacks a standard means of measuring a circuit’s potential for early termination and predicting at what cycle it would be safe to terminate. To fill this gap, we propose normalized stability, a metric that measures how fast a bit stream converges under a given accuracy budget. Our unit-level experiments show that normalized stability accurately reflects and contrasts the early-termination capabilities of varying stochastic computing units. Furthermore, our application-level experiments on low-density parity-check decoding, machine learning and image processing show that normalized stability can reduce the design space and predict the timing to terminate early.

关键词： Measurement Current measurement Streaming media Circuit stability Timing Velocity measurement Standards

来源：评论

学校读者我要写书评

暂无评论

A Forecast of Coronary Heart Disease using Proficient Machine Learning Algorithms

A Forecast of Coronary Heart Disease using Proficient Machin...

引用

International conference on image Information processing

作者： Shivani Gaba Alankrita Aggarwal Shally Nagpal Deepak Kumar Pardeep Singh Computer Science and Engineering Panipat Institute of Engineering and Technology Panipat Haryana India Cybernatics School of Computer Science University of Petroleum and Energy Studies Dehradun

ISBN: (纸本)9781665433624

Coronary Heart Diseases (CHDs) are a fundamental explanation of enormous deaths on earth in the last decades and are a dangerous disease in India and worldwide and Coronary Heart Disease has developed as one of the most unmistakable and uninformed reasons for death all around the globe. Thus, a dependable, precise & achievable framework for analyzing these maladies for appropriate therapy. Artificial Intelligence evaluations & systems are being used to restore data collections to robotize investigation within enormous & uneasy information. Numerous scientists, as of late, have been utilizing a few Artificial Intelligence methods to facilitate well-being for industry & professionals analysis of coronary-disease infections. This work intends to make use of chronological medical data to forecast CHD using Machine Learning. The work introduces machine learning techniques of different models dependent on calculations, procedures, and analyzes exhibition. Also, in this paper three supervised learning methods: Linear Regression using stochastic gradient descent and Decision Tree to find out the relationship in CHD data to improve prediction rate.

关键词： Heart Machine learning algorithms Service robots Supervised learning Linear regression stochastic processes Medical treatment

来源：评论

学校读者我要写书评

暂无评论

stochastic Optimization with Laggard Data Pipelines 34

Stochastic Optimization with Laggard Data Pipelines

引用

34th conference on Neural Information processing Systems (NeurIPS)

作者： Agarwal, Naman Anil, Rohan Koren, Tomer Talwar, Kunal Zhang, Cyril Google AI Princeton Princeton NJ 08540 USA Google Res Mountain View CA 94043 USA Tel Aviv Univ Tel Aviv Israel Google Tel Aviv Israel Apple Cupertino CA 95014 USA Microsoft Res New York NY 10012 USA

ISBN: (纸本)9781713829546

State-of-the-art optimization is steadily shifting towards massively parallel pipelines with extremely large batch sizes. As a consequence, CPU-bound preprocessing and disk/memory/network operations have emerged as new performance bottlenecks, as opposed to hardware-accelerated gradient computations. In this regime, a recently proposed approach is data echoing (Choi et al., 2019), which takes repeated gradient steps on the same batch while waiting for fresh data to arrive from upstream. We provide the first convergence analyses of "data-echoed" extensions of common optimization methods, showing that they exhibit provable improvements over their synchronous counterparts. Specifically, we show that in convex optimization with stochastic minibatches, data echoing affords speedups on the curvature-dominated part of the convergence rate, while maintaining the optimal statistical rate.

关键词： Convex optimization

来源：评论

学校读者我要写书评

暂无评论

Hyperspectral Anomaly Detection based on Frequency Analysis of Repeated Spatial Patterns 1

Hyperspectral Anomaly Detection based on Frequency Analysis ...

引用

1st International conference on Machine Vision and image processing, MVIP 2020

作者： Taghipour, Ashkan Ghassemian, Hassan Tarbiat Modares University Image Processing and Information Analysis Lab Faculty of Electrical and Computer Engineering Tehran Iran

ISBN: (纸本)9781728168326

Hyperspectral anomaly detection has brought significant attention to researchers during the last decade. In the remote sensing domain, it has lots of applications like environmental monitoring, mine source detection, human safety and, etc. Different methods have been proposed to overcome the challenges of anomaly detection, which mainly are a highly accurate detection rate and also having low false alarm rates. Background complexity of the hyperspectral data causes false detection of pixels as anomalies, and several statistical and sparse based methods try to solve it. However, the computation burden of the statistical and sparse based methods make them unpractical. Also, they often do not utilize the spatial information of hyperspectral data for increasing the detection accuracy. In this paper, we proposed a frequency analysis based approach that suppresses the repeatable patterns of the scene which mainly represent the background and considers the remainder regions as anomalies. Owe to the Fourier analysis advantage, it fast and accurate detects anomalous regions. Two well-known data sets have been chosen to challenge the potency of the proposed method, and the results are compared with some state-of-the-art methods. The visual and quantitative results confirm the supremacy of the proposed method to competitors in computation complexity and detection accuracy point of view. © 2020 IEEE.

关键词： Anomaly detection

来源：评论

学校读者我要写书评

暂无评论

Improving Inference for Neural image Compression 34

Improving Inference for Neural Image Compression

引用

34th conference on Neural Information processing Systems (NeurIPS)

作者： Yang, Yibo Bamler, Robert Mandt, Stephan Univ Calif Irvine Dept Comp Sci Irvine CA 92697 USA

ISBN: (纸本)9781713829546

We consider the problem of lossy image compression with deep latent variable models. State-of-the-art methods [Balle et al., 2018, Minnen et al., 2018, Lee et al., 2019] build on hierarchical variational autoencoders (VAEs) and learn inference networks to predict a compressible latent representation of each data point. Drawing on the variational inference perspective on compression [Alemi et al., 2018], we identify three approximation gaps which limit performance in the conventional approach: an amortization gap, a discretization gap, and a marginalization gap. We propose remedies for each of these three limitations based on ideas related to iterative inference, stochastic annealing for discrete optimization, and bits-back coding, resulting in the first application of bits-back coding to lossy compression. In our experiments, which include extensive baseline comparisons and ablation studies, we achieve new state-of-the-art performance on lossy image compression using an established VAE architecture, by changing only the inference method.

关键词： image compression

来源：评论

学校读者我要写书评

暂无评论

Maximum Likelihood Estimation of Regularization Parameters in High-Dimensional Inverse Problems: An Empirical Bayesian Approach Part I: Methodology and Experiments

引用

SIAM JOURNAL ON IMAGING SCIENCES 2020年第4期13卷 1945-1989页

作者： Vidal, Ana Fernandez De Bortoli, Valentin Pereyra, Marcelo Durmus, Alain Heriot Watt Univ Maxwell Inst Math Sci Edinburgh EH14 4AS Midlothian Scotland Heriot Watt Univ Sch Math & Comp Sci Edinburgh EH14 4AS Midlothian Scotland Univ Paris Saclay CNRS CMLA Ecole Normale Super Paris Saclay F-94235 Cachan France

Many imaging problems require solving an inverse problem that is ill-conditioned or ill-posed. Imaging methods typically address this difficulty by regularizing the estimation problem to make it well-posed. This often requires setting the value of the so-called regularization parameters that control the amount of regularization enforced. These parameters are notoriously difficult to set a priori and can have a dramatic impact on the recovered estimates. In this work, we propose a general empirical Bayesian method for setting regularization parameters in imaging problems that are convex w.r.t. the unknown image. Our method calibrates regularization parameters directly from the observed data by maximum marginal likelihood estimation and can simultaneously estimate multiple regularization parameters. Furthermore, the proposed algorithm uses the same basic operators as proximal optimization algorithms, namely gradient and proximal operators, and it is therefore straightforward to apply to problems that are currently solved by using proximal optimization techniques. Our methodology is demonstrated with a range of experiments and comparisons with alternative approaches from the literature. The considered experiments include image denoising, nonblind image deconvolution, and hyperspectral unmixing, using synthesis and analysis priors involving the Li, total-variation, total-variation and 4, and total-generalized-variation pseudonorms. A detailed theoretical analysis of the proposed method is presented in our companion paper [V. De Bortoli et al., SIAM J. Imaging Sci., 13 (2020), pp. 1990-2028].

关键词： image processing inverse problems statistical inference empirical Bayes stochastic optimization Markov chain Monte Carlo methods proximal algorithms

来源：评论

学校读者我要写书评

暂无评论

Contrast Enhancement of Medical images Using statistical methods with image processing Concepts 6

Contrast Enhancement of Medical Images Using Statistical Met...

引用

6th International Engineering conference "Sustainable Technology and Development", IEC 2020

作者： Al-Ameen, Zohair College of Computer Science and Mathematics University of Mosul Department of Computer Science Mosul-Nineveh Iraq

ISBN: (纸本)9781728159102

The process of generating pictorial representations of the inner human body for medical analysis and intervention is called medical imaging (MI). MI strives for revealing inner structures concealed by the existence of bones and skin to identify diseases in order to be treated properly. MI can be achieved through the use of various imaging systems, where each has different technology necessities. The output of these imaging systems is digital images that have meaningful medical information. Such images, however, do not own a high perceived quality due to the existence of image degradations. Despite modern technology advancement, various medical systems are still producing images with poor contrast owing to incorrect settings and device limitations. Such images must be processed efficiently to become clearer for better analysis, understanding and interpretation. In this study, a simple algorithm is developed, where it utilizes a combination of image processing concepts and statistical methods. The developed algorithm is appraised with a dataset of real-degraded low-contrast images only, assessed with one specialized no-reference metric and compared with four known contrast enhancement methods. From the conducted experiments, the proposed algorithm showed promising performances, since it produced acceptable-quality results rapidly and outperformed the comparison methods in different important aspects. © 2020 IEEE.

关键词： Medical imaging

来源：评论

学校读者我要写书评

暂无评论

A Low Latency stochastic Square Root Circuit

A Low Latency Stochastic Square Root Circuit

引用

International conference on VLSI Design

作者： Shyamali Mitra Debojyoti Banerjee Mrinal K. Naskar Jadavpur University Kolkata India

stochastic computing (SC) has shown promise over years due to its relative simplicity in hardware design enabling low power compared to conventional computing paradigm. The error tolerance and hardware-efficient solutions of SC have made it extremely popular in a variety of applications like decoding of Low Discrepancy sequences, image processing tasks, etc. In the present work, two architectures of stochastic Square Root Circuit (SSRC) have been proposed. The first architecture uses different constant-length stochastic numbers (SNs) and calculates the square root in two iterations providing reasonably accurate results throughout the input range. The accuracy is compared with state-of-the-art architectures and significant improvement in the accuracy for a lower range of inputs, i.e. 0-0.2 is noted. Another architecture consists of an accuracy control where the desired accuracy can be decided by a user. Studies are conducted to reveal that, in contrary to the fixed-length bitstream, a progressive increment of bitstream at each iteration utilizes a lesser number of clock cycles to converge to the desired accuracy for higher input range 0.6-1. The proposed SSRC is validated on two standard images for square root based contrast enhancement function to reveal close proximity with the traditional square root based enhanced image compared to other existing SSRCs.

关键词： image processing Computer architecture Very large scale integration Hardware Real-time systems Iterative methods Task analysis

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：