检索结果-内蒙古大学图书馆

44th IEEE International Conference on Acoustics, Speech and signal processing (ICASSP)

作者： Yu, Jianwei Lam, Max. W. Y. Chen, Xie Hu, Shoukang Liu, Songxiang Wu, Xixin Liu, Xunying Meng, Helen Chinese Univ Hong Kong Hong Kong Peoples R China Microsoft AI & Res One Microsoft Way Redmond WA USA

ISBN: (纸本)9781479981311

Recurrent neural network language models (RNNLMs) have become an increasing popular choice for state-of-the-art speech recognition systems. RNNLMs are normally trained by minimizing the cross entropy (CE) using the stochastic gradient descent (SGD) algorithm. However, the SGD method doesn't consider the correlation between parameters and therefore can lead to unstable and slow convergence in training. Second-order optimization methods provide a possible solution to this issue. However these methods are either computationally heavy or do not have competitive performance. In this paper, a novel optimization method - stochastic natural gradient based on minimum variance assumption (SNGM) is proposed for training RNNLMs. It allows the natural gradient method to operate at a comparable training efficiency to the SGD method. By modifying the gradient according to the local curvature of the KL-divergence between current and updated probabilistic distributions, the proposed SNGM approach is shown to outperform both the SGD and limited memory BFGS methods across three tasks: Penn Treebank, Switchboard conversational speech recognition and AMI meeting room transcription in terms of both perplexity and word error rate.

关键词： RNNLMs Natural Gradient

来源：评论

学校读者我要写书评

暂无评论

Roman domination-based spiking neural network for optimized EEG signal classification of four class motor imagery

引用

Computers in Biology and Medicine 2025年 194卷

作者： Raja Sekhar Banovoth Kadambari K V Department of Computer Science and Engineering National Institute of Technology Warangal Telangana 506004 India

The Spiking neural Network (SNN) is a third-generation neural network recognized for its energy efficiency and ability to process spatiotemporal information, closely imitating the behavioral mechanisms of biological neurons in the brain. SNN exhibit rich neurodynamic features in the spatiotemporal domain, making them well-suited for processing brain signals, mainly those captured using the widely used non-invasive Electroencephalography (EEG) technique. However, the structural limitations of SNN hinder their feature extraction capabilities for motor imagery signal classification, which leads to under performance of the task. To address the aforementioned challenge, the proposed study introduces a novel model that incorporates Roman Domination within a Spiking neural Network (RDSNN), where Roman domination identifies the most highly correlated channels or nodes. These channels generate an appropriate threshold for spike generation in the signals, which are then classified using the SNN. The model’s performance was evaluated on three typically representative motor imagery datasets: PhysioNet, BCI Competition IV-2a, and BCI Competition IV-2b. RDSNN achieved 73.65% accuracy on PhysioNet, 81.75% on BCI IV-2a, and 84.56% on BCI IV-2b. The results demonstrate not only superior accuracy compared to State-Of-the-Art (SOTA) methods but also a 35% reduction in computation time, attributed to the application of Roman domination.

关键词： image correlation

来源：评论

学校读者我要写书评

暂无评论

image CORRECTION IN EMISSION TOMOGRAPHY USING DEEP CONVOLUTION neural NETWORK 44

IMAGE CORRECTION IN EMISSION TOMOGRAPHY USING DEEP CONVOLUTI...

引用

44th IEEE International Conference on Acoustics, Speech and signal processing (ICASSP)

作者： Suzuki, Tomohiro Kudo, Hiroyuki Univ Tsukuba Grad Sch Syst & Informat Engn Dept Comp Sci Tennoudai 1-1-1 Tsukuba Ibaraki 3058573 Japan

ISBN: (纸本)9781479981311

We propose a new approach using Deep Convolution neural Network (DCNN) to correct for image degradations due to statistical noise and photon attenuation in Emission Tomography (ET). The proposed approach first reconstructs an image by the standard Filtered Backprojection (FBP) without correcting for the degradations followed by inputting the degraded image into DCNN to obtain an improved image. We consider two different scenarios. The first scenario inputs an ET image only into DCNN, whereas the second scenario inputs a pair of degraded ET image and CT/MRI image to improve accuracy of the correction. The simulation result demonstrates that both the scenarios can improve image quality compared to the FBP without correction, and, in particular, accuracy of the second scenario is comparable to that of the standard iterative reconstruction such as Maximum Likelihood Expectation Maximization (MLEM) and Ordered-Subsets EM (OSEM) methods. The proposed method is able to output an image in very short time, because it does not rely on iterative computations.

关键词： Convolution neural network emission tomography image reconstruction attenuation correction

来源：评论

学校读者我要写书评

暂无评论

CONVOLUTIONAL neural NETWORKS FOR HETEROGENEOUS INGREDIENT DISCRIMINATION WITH HYPERSPECTRAL IMAGING 10

CONVOLUTIONAL NEURAL NETWORKS FOR HETEROGENEOUS INGREDIENT D...

引用

10th Workshop on Hyperspectral Imaging and signal processing: Evolution in Remote Sensing (WHISPERS)

作者： Blanch-Perez-del-Notario, Carolina Saeys, Wouter Lambrechts, Andy IMEC Kapeldreef 75 B-3001 Leuven Belgium Katholieke Univ Leuven Div Mechatron Biostat & Sensors B-3001 Leuven Belgium

ISBN: (纸本)9781728152943

Convolutional neural Networks (CNNs) are recently gaining popularity to perform a joint spatio-spectral analysis of hyperspectral images and have achieved good performance in remote sensing applications. We show the potential of CNNs for an industrial application of heterogeneous ingredient detection and show a significant discrimination gain with respect to traditional machine learning methods. Additionally, we explore the potential of using downsampled spatio-spectral resolutions of the hyperspectral image achieving high discrimination while reducing data storage, acquisition and computational requirements. Finally, we show how CNNs can enable the use of low-resolution snapshot cameras, which allow portability and fast acquisition in industrial applications.

关键词： Convolutional neural network spatio-spectral resolution ingredient identification

来源：评论

学校读者我要写书评

暂无评论

KR product and sparse prior based CNN estimator for 2-D DOA estimation

引用

AEU-INTERNATIONAL JOURNAL OF ELECTRONICS AND COMMUNICATIONS 2021年 137卷 153780-153780页

作者： Yuan, Ye Wu, Shuang Ma, Yuhong Huang, Lei Yuan, Naichang Natl Univ Def Technol State Key Lab Complex Electromagnet Environm Effe Deya Rd 109 Changsha 410073 Hunan Peoples R China Acad Mil Sci Natl Innovat Inst Def Technol Fengtai West Rd Beijing 100071 Peoples R China

This paper proposes a method based on Khatri-Rao (KR) product, sparse prior, and convolutional neural networks (CNN) to solve the direction-of-arrival (DOA) estimation problem. Firstly, we use the KR product to expand the degree of freedom (DOF) of the 2-D antenna array. Then we calculate the sparse power spectrum of signals and obtain an RGB image tensor of the spectrum. Finally, we design a CNN group with three different sub-networks to estimate 2-D DOA information. Two of the sub-networks are used for obtaining the spectrum of azimuth angle and elevation angle, respectively. One specific network is designed as the pairing network used for paring azimuth angle with the correct elevation angle. The proposed CNN group is data-driven and does not rely on any prior knowledge of incidence signals. We investigate the feature of estimation error, the root mean squared error (RMSE) responses under different experiment environments, the resolution of the proposed estimation CNN group, and the pairing performance of the proposed pairing network. Comparing with prior estimation methods, the proposed CNN group shows satisfactory estimation accuracy and stability.

关键词： Array signal processing Artificial intelligence Convolutional neural network Direction of arrival estimation Supervised learning

来源：评论

学校读者我要写书评

暂无评论

时频域重叠多信号智能检测方法研究

引用

信号处理 2021年第5期37卷 878-884页

作者：李杰孙闽红仇兆炀杭州电子科技大学通信工程学院浙江杭州310018

针对现有基于深度学习理论的信号智能检测方法大多只能对单信号或时频域不重叠的信号进行检测,本文提出了一种基于掩膜区域卷积神经网络(Mask R-CNN)与Criminisi算法的时频重叠多信号智能检测新方法。首先将一维时域信号通过时频变换得... 详细信息

针对现有基于深度学习理论的信号智能检测方法大多只能对单信号或时频域不重叠的信号进行检测,本文提出了一种基于掩膜区域卷积神经网络(Mask R-CNN)与Criminisi算法的时频重叠多信号智能检测新方法。首先将一维时域信号通过时频变换得到二维时频图像。然后针对时频图中多信号重叠部分像素位置信息缺失这一问题,提出了利用Criminisi算法对信号重叠部分像素位置信息进行恢复。最后,基于缺失信息恢复后的图像使用Mask R-CNN进行训练,再用训练后的网络对未知信号进行检测。实验结果表明,该方法在信噪比(SNR)为-3 dB时,时频域重叠信号的平均检测率达92%,相比基于卷积神经网络的信号检测方法,在SNR大于-3 dB时检测率平均提高20%以上。

关键词：信号检测卷积神经网络时频重叠 Criminisi算法

来源：评论

学校读者我要写书评

暂无评论

Data, signal and image processing and Applications in Sensors

引用

SENSORS 2021年第10期21卷 3323-3323页

作者： Reis, Manuel J. C. S. Univ Tras Os Montes & Alto Douro UTAD IEETA Dept Engn P-5000801 Vila Real Portugal

In order to obtain relevant and insightful metrics from the sensors signals’ data, further enhancement of the acquired sensor signals, such as the noise reduction in the one-dimensional electroencephalographic (EEG) signals or color correction in the endoscopic images, and their analysis by computer-based medical systems, is needed. The proposed SER model was evaluated over two benchmarks, which included the interactive emotional dyadic motion capture (IEMOCAP) and the berlin emotional speech database (EMO-DB) speech datasets, and it obtained 77.01% and 92.02% recognition results, showing a better recognition performance than the state-of-the-art SER systems. [5] proposed a demodulation method based on Loran-C Pulse Envelope Correlation–Phase Detection (EC–PD), in which EC has two implementation schemes, namely, moving average-cross correlation and matched correlation, to reduce the effects of noise and SkyWave Interference (SWI). The experimental results, on the public GoPro dataset and the realistic and dynamic scenes (REDS) dataset, show that the proposed method generally outperforms some traditional deburring methods and deep-learning-based, state-of-the-art deblurring methods, such as scale-recurrent network (SRN) and denoising prior driven deep neural network (DPDNN), in terms of such quantitative indexes as peak signal-to-noise ratio (PSNR) and structural similarity (SSIM) and human vision.

关键词：

来源：评论

学校读者我要写书评

暂无评论

methods to enhance the automation of operational modal analysis 45

Methods to enhance the automation of operational modal analy...

引用

45th International Conference on Vibroengineering

作者： Wiemann, Marcel Bonekemper, Lukas Kraemer, Peter Department of Mechanical Engineering University of Siegen Siegen Germany

The vibration-based damage detection and the monitoring of modal data are currently based on different Operational Modal Analysis (OMA) approaches. For the continuous monitoring of modal quantities, different techniques for automated feature extraction are known. Especially in recent years several research groups and companies have been working on the automatic interpretation of stability plots. Nevertheless, many questions regarding data pre-processing for OMA in time or frequency domain are still unanswered. The present paper deals with issues regarding effective pre-processing methods for OMA based on Covariance-stochastic Subspace Identification. In this context, the orthogonality of matrices after model order reduction, etc. are referred. This includes, for example, a comparison between the classical calculation of the reduced-order matrices and a procedure that preserves the orthogonality of these matrices. A method known from the signal denoising and image processing is also successful used to extract and select the modes. The mode extraction method is validated with an innovative three-dimensional stability plot. This paper does not claim to solve all tasks of an automated OMA, but it contributes the calculation of clean, easy to interpret, stability plots, which should facilitate the automatic evaluation in the future. The effectiveness of the algorithms is demonstrated by means of simulated (3DOF-StateSpace) and measured data of a laboratory structure described in [1]. Afterwards the results and the future works on the topic are discussed. Copyright © 2020 Marcel Wiemann, et al. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

关键词： Structural health monitoring

来源：评论

学校读者我要写书评

暂无评论

CONVOLUTIONAL neural NETWORKS CONSIDERING LOCAL AND GLOBAL FEATURES FOR image ENHANCEMENT 26

CONVOLUTIONAL NEURAL NETWORKS CONSIDERING LOCAL AND GLOBAL F...

引用

26th IEEE International Conference on image processing (ICIP)

作者： Kinoshita, Yuma Kiya, Hitoshi Tokyo Metropolitan Univ Tokyo Japan

ISBN: (纸本)9781538662496

In this paper, we propose a novel convolutional neural network (CNN) architecture considering both local and global features for image enhancement. Most conventional image enhancement methods, including Retinex-based methods, cannot restore lost pixel values caused by clipping and quantizing. CNN-based methods have recently been proposed to solve the problem, but they still have a limited performance due to network architectures not handling global features. To handle both local and global features, the proposed architecture consists of three networks: a local encoder, a global encoder, and a decoder. In addition, high dynamic range (HDR) images are used for generating training data for our networks. The use of HDR images makes it possible to train CNNs with better-quality images than images directly captured with cameras. Experimental results show that the proposed method can produce higher-quality images than conventional image enhancement methods including CNN-based methods, in terms of various objective quality metrics: TMQI, entropy, NIQE, and BRISQUE.

关键词： image enhancement High dynamic range images Deep learning Convolutional neural networks

来源：评论

学校读者我要写书评

暂无评论

BigNeuron: a resource to benchmark and predict performance of algorithms for automated tracing of neurons in light microscopy datasets

引用

Nature methods 2023年第6期20卷 824-835页

作者： Linus Manubens-Gil Zhi Zhou Hanbo Chen Arvind Ramanathan Xiaoxiao Liu Yufeng Liu Alessandro Bria Todd Gillette Zongcai Ruan Jian Yang Miroslav Radojević Ting Zhao Li Cheng Lei Qu Siqi Liu Kristofer E Bouchard Lin Gu Weidong Cai Shuiwang Ji Badrinath Roysam Ching-Wei Wang Hongchuan Yu Amos Sironi Daniel Maxim Iascone Jie Zhou Erhan Bas Eduardo Conde-Sousa Paulo Aguiar Xiang Li Yujie Li Sumit Nanda Yuan Wang Leila Muresan Pascal Fua Bing Ye Hai-Yan He Jochen F Staiger Manuel Peter Daniel N Cox Michel Simonneau Marcel Oberlaender Gregory Jefferis Kei Ito Paloma Gonzalez-Bellido Jinhyun Kim Edwin Rubel Hollis T Cline Hongkui Zeng Aljoscha Nern Ann-Shyn Chiang Jianhua Yao Jane Roskams Rick Livesey Janine Stevens Tianming Liu Chinh Dang Yike Guo Ning Zhong Georgia Tourassi Sean Hill Michael Hawrylycz Christof Koch Erik Meijering Giorgio A Ascoli Hanchuan Peng Institute for Brain and Intelligence Southeast University Nanjing China. Microsoft Corporation Redmond WA USA. Tencent AI Lab Bellevue WA USA. Computing Environment and Life Sciences Directorate Argonne National Laboratory Lemont IL USA. Kaya Medical Seattle WA USA. University of Cassino and Southern Lazio Cassino Italy. Center for Neural Informatics Structures and Plasticity Krasnow Institute for Advanced Study George Mason University Fairfax VA USA. Faculty of Information Technology Beijing University of Technology Beijing China. Beijing International Collaboration Base on Brain Informatics and Wisdom Services Beijing China. Nuctech Netherlands Rotterdam the Netherlands. Janelia Research Campus Howard Hughes Medical Institute Ashburn VA USA. Department of Electrical and Computer Engineering University of Alberta Edmonton Alberta Canada. Ministry of Education Key Laboratory of Intelligent Computation and Signal Processing Anhui University Hefei China. Paige AI New York NY USA. Scientific Data Division and Biological Systems and Engineering Division Lawrence Berkeley National Lab Berkeley CA USA. Helen Wills Neuroscience Institute and Redwood Center for Theoretical Neuroscience UC Berkeley Berkeley CA USA. RIKEN AIP Tokyo Japan. Research Center for Advanced Science and Technology (RCAST) The University of Tokyo Tokyo Japan. School of Computer Science University of Sydney Sydney New South Wales Australia. Texas A&M University College Station TX USA. Cullen College of Engineering University of Houston Houston TX USA. Graduate Institute of Biomedical Engineering National Taiwan University of Science and Technology Taipei Taiwan. National Centre for Computer Animation Bournemouth University Poole UK. PROPHESEE Paris France. Department of Neuroscience Columbia University New York NY USA. Mortimer B. Zuckerman Mind Brain Behavior Institute Columbia University New York NY USA. Department of Computer Science Northern Illinois Universit

BigNeuron is an open community bench-testing platform with the goal of setting open standards for accurate and fast automatic neuron tracing. We gathered a diverse set of image volumes across several species that is representative of the data obtained in many neuroscience laboratories interested in neuron tracing. Here, we report generated gold standard manual annotations for a subset of the available imaging datasets and quantified tracing quality for 35 automatic tracing algorithms. The goal of generating such a hand-curated diverse dataset is to advance the development of tracing algorithms and enable generalizable benchmarking. Together with image quality features, we pooled the data in an interactive web application that enables users and developers to perform principal component analysis, t-distributed stochastic neighbor embedding, correlation and clustering, visualization of imaging and tracing data, and benchmarking of automatic tracing algorithms in user-defined data subsets. The image quality metrics explain most of the variance in the data, followed by neuromorphological features related to neuron size. We observed that diverse algorithms can provide complementary information to obtain accurate results and developed a method to iteratively combine methods and generate consensus reconstructions. The consensus trees obtained provide estimates of the neuron structure ground truth that typically outperform single algorithms in noisy datasets. However, specific algorithms may outperform the consensus tree strategy in specific imaging conditions. Finally, to aid users in predicting the most accurate automatic tracing results without manual annotations for comparison, we used support vector machine regression to predict reconstruction quality given an image volume and a set of automatic tracings.

关键词：

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：