检索结果-内蒙古大学图书馆

IEEE International Conference on Acoustics, Speech and Signal Processing

作者： J. Hahn A. M. Zoubir Signal Process. Group Tech. Univ. Darmstadt Darmstadt Germany

ISBN: (纸本)9781467369985

Reinforcement Learning (RL) is an attractive tool for learning optimal controllers in the sense of a given reward function. In conventional RL, usually an expert is required to design the reward function as the efficiency of RL strongly depends on the latter. An alternative has been presented by the concept of Inverse Reinforcement Learning (IRL), where the reward function is estimated from observed data. In this work, we propose a novel approach for IRL based on a generative probabilistic model of RL. We derive an expectation Maximization algorithm that is able to simultaneously estimate the reward and the optimal policy for finite state and action spaces, which can be easily extended for the infinite cases. By means of two toy examples, we show that the proposed algorithm works well even with a low number of observations and converges after only a few iterations.

关键词： expectation Maximization Inverse Reinforcement Learning Markov Decision Process Markov chain learning (artificial intelligence) Reward expectation-maximisation algorithm IRL Learning Probabilistic Model Inverse optimal strategy

来源：评论

学校读者我要写书评

暂无评论

An innovative multimodal/multispectral image registration method for medical images based on the expectation-maximization algorithm

An innovative multimodal/multispectral image registration me...

引用

Annual International Conference of the IEEE Engineering in Medicine and Biology Society

作者： Edgar Arce-Santana Daniel U. Campos-Delgado Aldo Mejia-Rodriguez Isnardo Reducindo Facultad de Ciencias Universidad Autonoma de San Luis Potosi SLP Mexico

ISBN: (纸本)9781424492695

In this paper, we present a methodology for multimodal/multispectral image registration of medical images. This approach is formulated by using the expectation-Maximization (EM) methodology, such that we estimate the parameters of a geometric transformation that aligns multimodal/multispectral images. In this framework, the hidden random variables are associated to the intensity relations between the studied images, which allow to compare multispectral intensity values between images of different modalities. The methodology is basically composed by an iterative two-step procedure, where at each step, a new estimation of the joint conditional multispectral intensity distribution and the geometric transformation is computed. The proposed algorithm was tested with different kinds of medical images, and the obtained results show that the proposed methodology can be used to efficiently align multimodal/multispectral medical images.

关键词： EM medical images based expectation-maximization algorithm expectation-maximisation algorithm image registration Geometric transformation images Medical imaging Medical intensity distribution

来源：评论

学校读者我要写书评

暂无评论

Distributed tls estimation under random data faults

Distributed tls estimation under random data faults

引用

IEEE International Conference on Acoustics, Speech and Signal Processing

作者： S. Silva Pereira A. Pages-Zamora R. Lopez-Valcarce SPCOM Group Univ. Politec. de Catalunya-Barcelona Tech Barcelona Spain

ISBN: (纸本)9781467369985

This paper addresses the problem of distributed estimation of a parameter vector in the presence of noisy input and noisy output data, as well as data faults, performed by a wireless sensor network in which only local interactions among the nodes are allowed. In the presence of unreliable observations, standard estimators become biased and perform poorly in low signal-to-noise ratios. We propose therefore two different distributed approaches based on the expectation-Maximization algorithm: in the first one the regressors are estimated at each iteration, whereas the second one does not require explicit regressor estimation. Numerical results show that the proposed methods approach the performance of a clairvoyant scheme with knowledge of the random data faults.

关键词： Diffusion distributed estimation expectation-maximization sensor networks total least squares random data Sensor networks faults Noise Least squares method Wireless sensor networks expectation-maximisation algorithm Disease Predictive Factor Distribute signal to noise ratios diffusion

来源：评论

学校读者我要写书评

暂无评论

Identification of dynamical biological systems based on random effects models

Identification of dynamical biological systems based on rand...

引用

Annual International Conference of the IEEE Engineering in Medicine and Biology Society

作者： Levy Batista Thierry Bastogne El-Hadi Djermoune CRAN CNRS UMR 7039 BP 70239 F-54506 Vandoeuvre-les-Nancy Cedex France

ISBN: (纸本)9781424492695

System identification is a data-driven modeling approach more and more used in biology and biomedicine. In this application context, each assay is always repeated to estimate the response variability. The inference of the modeling conclusions to the whole population requires to account for the inter-individual variability within the modeling procedure. One solution consists in using random effects models but up to now no similar approach exists in the field of dynamical system identification. In this article, we propose a new solution based on an ARX (Auto Regressive model with eXternal inputs) structure using the EM (expectation-maximisation) algorithm for the estimation of the model parameters. Simulations show the relevance of this solution compared with a classical procedure of system identification repeated for each subject.

关键词： ARX EM Identification ARX gene random effect System identification Response Variability biomedicine biological organs Authentication expectation-maximisation algorithm Solutions

来源：评论

学校读者我要写书评

暂无评论

A vessel segmentation method for MRA data based on probabilistic mixture model

A vessel segmentation method for MRA data based on probabili...

引用

2015 IET International Conference on Biomedical Image and Signal Processing (ICBISP 2015)

作者： Pei Lu Cheng Wang Shoujun Zhou Shenzhen Institutes of Advanced Technology Chinese Academy of Sciences Shenzhen China

Accurate segmentation and visualization of cerebral vessels have important significance to the diagnosis and treatment of relevant brain diseases. However, some segmentation algorithms are only competent for the standard data. In this paper, a segmentation method based on probabilistic mixture model is proposed to solve clinical problems. Through histogram analysis of the magnetic resonance angiography (MRA) data offered by Guangzhou General Hospital of the Chinese PLA, a mixture model formed by six probabilistic distributions (one Exponential, one Rayleigh, and four Normal distributions) was built to fit the histogram curve. Least squares and expectation maximization (EM) algorithms have been used for parameters estimation. At last, the segmentation was enhanced by maximum a posteriori probability (MAP) and Markov random field (MRF) algorithm. The effectiveness of the proposed method has been validated by segmentation tests on a series of clinical MRA data with good performance.

关键词： EM algorithm MRF Vessel segmentation histogram analysis mixture model Mixture models expectation-maximisation algorithm Probability histograms MYRF gene Maximum a posteriori estimation Parameter estimation cerebrovascular system

来源：评论

学校读者我要写书评

暂无评论

Hessian-free Optimization for Learning Deep Multidimensional Recurrent Neural Networks 15

Hessian-free Optimization for Learning Deep Multidimensional...

引用

Annual Conference on Neural Information Processing Systems

作者： Minhyung Cho Chandra Shekhar Dhir Jaehyung Lee Applied Research Korea Gracenote Inc.

ISBN: (纸本)9781510825024

Multidimensional recurrent neural networks (MDRNNs) have shown a remarkable performance in the area of speech and handwriting recognition. The performance of an MDRNN is improved by further increasing its depth, and the difficulty of learning the deeper network is overcome by using Hessian-free (HF) optimization. Given that connectionist temporal classification (CTC) is utilized as an objective of learning an MDRNN for sequence labeling, the non-convexity of CTC poses a problem when applying HF to the network. As a solution, a convex approximation of CTC is formulated and its relationship with the EM algorithm and the Fisher information matrix is discussed. An MDRNN up to a depth of 15 layers is successfully trained using HF, resulting in an improved performance for sequence labeling.

关键词： recurrent neural nets expectation-maximisation algorithm Hf(hafnium) MULTIDIMENSIONAL High frequency information matrix Speech handwriting recognition

来源：评论

学校读者我要写书评

暂无评论

Hidden Markov model parameters estimation with independent multiple observations and inequality constraints

引用

IET SIGNAL PROCESSING 2014年第9期8卷 938-949页

作者： Xia, Lisha Fang, Huajing Liu, Xiaoyong Huazhong Univ Sci & Technol Sch Automat Wuhan 430074 Peoples R China

In this study, the authors focus on hidden Markov model (HMM) parameters estimation with independent multiple observations and non-linear inequality constraints. The parameters estimation process is divided into four steps: initialisation, parameters pre-estimation, parameters re-estimation and termination. The pre-estimation results are used to approximate non-linear inequality constraints to linear inequality constraints. In parameters re-estimation step, the active-set optimisation is combined with the expectation maximisation (EM) algorithm in M-step and the active set-based EM algorithm is proposed to re-estimate HMM parameters when inequality constraints are not satisfied in pre-estimation. An auxiliary function is devised for reconstructing the optimisation objective function and the convergence of the proposed algorithm is also demonstrated. Simulation results indicate that the proposed algorithm provides better performance by modifying the random error of observation data appropriately and it is powerful for industry process fault diagnosis.

关键词： hidden Markov models parameter estimation nonlinear estimation approximation theory expectation-maximisation algorithm optimisation fault diagnosis hidden Markov model parameter estimation independent multiple observation HMM parameter estimation parameters pre-estimation step parameters reestimation step termination step initialisation step approximate nonlinear inequality constraint linear inequality constraint active-set optimisation expectation maximisation algorithm M-step algorithm active set-based EM algorithm optimisation objective function reconstruction random error modification industry process fault diagnosis

来源：评论

学校读者我要写书评

暂无评论

Unsupervised change detection in SAR images based on locally fitting model and semi-EM algorithm

引用

INTERNATIONAL JOURNAL OF REMOTE SENSING 2014年第2期35卷 621-650页

作者： Su, Linzhi Gong, Maoguo Sun, Bo Jiao, Licheng Xidian Univ Minist Educ Key Lab Intelligent Percept & Image Understanding Xian 710071 Peoples R China

Change detection in synthetic aperture radar (SAR) images has become increasingly important along with the development of SAR techniques. In this study, a novel locally fitting and expectation-maximization (EM) approach is proposed for unsupervised change detection tasks in SAR images. A difference image is generated first, and an in-depth study of the inherent characteristics of the histogram of the difference image is then made. Thus, the new approach is proposed corresponding to these characteristics. In this approach, the locally fitting model orientated to deal with the unchanged class is put forward to reach a best fit, and the semi-EM algorithm for the changed class is used to tackle the phenomenon of overlapping. Then, through the Bayesian decision rule, the optimal threshold is determined. Our contributions lie in two aspects. First, in the locally fitting model, the location of the optimal threshold can be determined, which leads to an accurate fit over a short interval. Second, the use of the semi-EM algorithm not only retains the efficacy of the EM algorithm to cope with overlapping for the changed class but also simplifies the computing process. Experiments on real data sets confirm the effectiveness of the proposed approach, which results in final maps very similar to the ground truth and is more effective in determining the optimal threshold in comparison with others. The experimental results also demonstrate its effectiveness when the changed areas are of different geometrical shapes.

关键词： expectation-maximisation algorithm ground truth Radar polarimetry difference image Specific absorption rate images Rescue mode matching

来源：评论

学校读者我要写书评

暂无评论

Joint phase noise estimation and data detection in coded multi-input-multi-output systems

引用

IET COMMUNICATIONS 2014年第7期8卷 981-989页

作者： Isikman, Arif O. Mehrpouyan, Hani Nasir, Ali A. Amat, Alexandre G. Kennedy, Rodney. A. Chalmers Univ Technol Commun Syst Grp S-41296 Gothenburg Sweden Calif State Univ Dept Elect & Comp Engn Bakersfield CA USA Australian Natl Univ Res Sch Engn Canberra ACT Australia

The problem of joint oscillator phase noise (PHN) estimation and data detection for multi-input multi-output (MIMO) systems using bit-interleaved-coded modulation is analysed. A new MIMO receiver that iterates between the estimator and the detector, based on the expectation-maximisation (EM) framework, is proposed. It is shown that at high signal-to-noise ratios, a maximum a posteriori (MAP) estimator can be used to carry out the maximisation step of the EM algorithm. Moreover, to reduce the computational complexity of the proposed EM algorithm, a soft decision-directed extended Kalman filter-smoother (EKFS) is applied instead of the MAP estimator to track the PHN parameters. The numerical results show that by combining the proposed EKFS-based approach with an iterative detector that employs low-density parity check codes, PHN can be accurately tracked. The simulations also demonstrate that compared to the existing algorithms, the proposed iterative receiver can significantly enhance the performance of MIMO systems in the presence of PHN.

关键词： phase noise MIMO communication radio receivers interleaved codes oscillators expectation-maximisation algorithm maximum likelihood estimation Kalman filters parity check codes iterative methods data detection coded multi-input-multi-output system joint oscillator phase noise estimation PHN estimation MIMO receiver bit-interleaved-coded modulation expectation-maximisation EM algorithm signal-to-noise ratio maximum a posteriori MAP estimator soft decision-directed extended Kalman filter-smoother EKFS-based approach low-density parity check code iterative receiver

来源：评论

学校读者我要写书评

暂无评论

Skew Gaussian mixture models for speaker recognition

引用

IET SIGNAL PROCESSING 2014年第8期8卷 860-867页

作者： Matza, Avi Bistritz, Yuval Tel Aviv Univ Sch Elect Engn IL-69978 Tel Aviv Israel

Gaussian mixture models (GMMs) are widely used in speech and speaker recognition. This study explores the idea that a mixture of skew Gaussians might capture better feature vectors that tend to have skew empirical distributions. It begins with deriving an expectation maximisation (EM) algorithm to train a mixture of two-piece skew Gaussians that turns out to be not much more complicated than the usual EM algorithm used to train symmetric GMMs. Next, the algorithm is used to compare skew and symmetric GMMs in some simple speaker recognition experiments that use Mel frequency cepstral coefficients (MFCC) and line spectral frequencies (LSF) as the feature vectors. MFCC are one of the most popular feature vectors in speech and speaker recognition applications. LSF were chosen because they exhibit significantly more skewed distribution than MFCC and because they are widely used [together with the related immittance spectral frequencies (ISF)] in speech transmission standards. In the reported experiments, models with skew Gaussians performed better than models with symmetric Gaussians and skew GMMs with LSF compared favourably with both skew symmetric and symmetric GMMs that used MFCC.

关键词： Gaussian processes mixture models speaker recognition vectors expectation-maximisation algorithm speaker recognition GMM speech recognition skew empirical distribution expectation maximisation algorithm EM algorithm two-piece skew Gaussian mixture model Mel frequency cepstral coefflcient MFCC line spectral frequency LSF immittance spectral frequency ISF speech transmission standard feature vectors

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：