Network tomography using multicast probes enables inference of loss characteristics of internal network links from reports of end-to-end loss seen at multicast receivers. In this paper, we develop estimators for inter...
详细信息
Network tomography using multicast probes enables inference of loss characteristics of internal network links from reports of end-to-end loss seen at multicast receivers. In this paper, we develop estimators for internal loss rates when reports are not available on all probes or from all receivers. This problem is motivated by the use of unreliable transport protocols, such as reliable transport protocol, to transmit loss reports to a collector for inference. We use a maximum-likelihood (NIL) approach in which we apply the expectationmaximization (EM) algorithm to provide an approximating solution to the the NIL estimator for the incomplete data problem. We present a concrete realization of the algorithm that can be applied to measured data. For classes of models. we establish identifiability of the probe and report loss parameters, and convergence of the EM sequence to the maximum-likelihood estimator (MLE). Numerical results suggest that these properties hold more generally. We derive convergence rates for the EM iterates, and the estimation error of the MLE. Finally, we evaluate the accuracy and convergence rate through extensive simulations.
The difficulty in tracking a maneuvering target in the presence of false measurements arises from the uncertain origin of the measurements (as a result of the observation detection process) and the uncertainty in the ...
详细信息
The difficulty in tracking a maneuvering target in the presence of false measurements arises from the uncertain origin of the measurements (as a result of the observation detection process) and the uncertainty in the maneuvering command driving the state of the target. Conditional mean estimates of the target state require a computational cost which is exponential with the number of observations and the levels of the maneuver command. In this paper, we propose an alternative optimal state estimation algorithm. Unlike the conditional mean estimator, which require computational cost exponential in the data length, the proposed iterative algorithm is linear in the data length (per iteration). The proposed iterative off-line algorithm optimally combines a hidden Markov model and a Kalman smoother-the optimality is demonstrated via the expectation maximization algorithm-to yield the maximum a posteriori trajectory estimate of the target state. The algorithm proposed in this paper, uses probabilistic multi-hypothesis (PMHT) techniques for tracking a single maneuvering target in clutter. The extension of our algorithm to multiple maneuvering target tracking is straightforward and details are omitted. Previous applications of the PMHT technique (IEEE Trans. Automat. Control, submitted) have addressed the problem of tracking multiple non-maneuvering targets. These techniques are extended to address the problem of optimal tracking of a maneuvering target in a cluttered environment. (C) 2002 Elsevier Science B.V. All rights reserved.
In this paper, finite-dimensional recursive filters for space-time Markov random fields are derived. These filters can be used with the expectationmaximization (EM) algorithm to yield maximum likelihood estimates of ...
详细信息
In this paper, finite-dimensional recursive filters for space-time Markov random fields are derived. These filters can be used with the expectationmaximization (EM) algorithm to yield maximum likelihood estimates of the parameters of the model. (C) 2002 Elsevier Science Ltd. All rights reserved.
作者:
Naphade, MRIBM Corp
Thomas J Watson Res Ctr Pervas Media Management Grp Hawthorne NY 10532 USA
Media analysis for video indexing is witnessing an increasing influence of statistical techniques. Examples of these techniques include the use of generative models as well as discriminant techniques for video structu...
详细信息
ISBN:
(纸本)0819446416
Media analysis for video indexing is witnessing an increasing influence of statistical techniques. Examples of these techniques include the use of generative models as well as discriminant techniques for video structuring, classification, summarization, indexing and retrieval. Advances in multimedia analysis are related directly to advances in signal processing, computer vision, pattern recognition, multimedia databases and smart sensors. This paper highlights the statistical techniques in multimedia retrieval with particular emphasis on semantic characterization.
The problem of estimating the parameters that determine a mixture density has been subject of extraordinary interest in the last years. Such a mixture density estimation problem is multi-target tracking. One of the ne...
详细信息
ISBN:
(纸本)0780376013
The problem of estimating the parameters that determine a mixture density has been subject of extraordinary interest in the last years. Such a mixture density estimation problem is multi-target tracking. One of the new data association/tracking algorithms used for multi-target tracking is Probabilistic Multi-Hypothesis Tracker proposed by Streit and Luginbuhl. This paper considers a modification of Probabilistic Multi-Hypothesis Tracker. The implementation of Hough Transform is proposed to generate a good start point to the main PMHT algorithm. The suggested modification improves estimation accuracy and correctness of PMHT algorithm and overcome its main drawback of low convergence.
This paper concerns the application of the EM algorithm for the estimation of the parameters of non-stationary noisy phase mono component signals buried in additive noise. Noisy phase signals are appropriate for model...
详细信息
ISBN:
(纸本)0780374029
This paper concerns the application of the EM algorithm for the estimation of the parameters of non-stationary noisy phase mono component signals buried in additive noise. Noisy phase signals are appropriate for modeling real world signals such as radar and communications signals. The maximum likelihood estimator for the signal parameters is explicited. The problem is then formulated as a missing data problem which enables a natural use of the expectation-maximizationalgorithm. A robust initialization scheme based on non-parametric time-frequency distributions is presented. Experimental results with both real and simulated data show the efficiency of the procedure.
The EM algorithm is a popular method for parameter estimation in a variety of problems involving missing data. However, the EM algorithm often requires significant computational resources and has been dismissed as imp...
详细信息
The EM algorithm is a popular method for parameter estimation in a variety of problems involving missing data. However, the EM algorithm often requires significant computational resources and has been dismissed as impractical for large databases. We present two approaches that significantly reduce the computational cost of applying the EM algorithm to databases with a large number of cases, including databases with large dimensionality. Both approaches are based on partial E-steps for which we can use the results of Neal and Hinton (In Jordan, M. (Ed.), Learning in Graphical Models, pp. 355-371. The Netherlands: Kluwer Academic Publishers) to obtain the standard convergence guarantees of EM. The first approach is a version of the incremental EM algorithm, described in Neal and Hinton (1998), which cycles through data cases in blocks. The number of cases in each block dramatically effects the efficiency of the algorithm. We provide a method for selecting a near optimal block size. The second approach, which we call lazy EM, will, at scheduled iterations, evaluate the significance of each data case and then proceed for several iterations actively using only the significant cases. We demonstrate that both methods can significantly reduce computational costs through their application to high-dimensional real-world and synthetic mixture modeling problems for large databases.
An unsupervised retraining technique for a maximum likelihood (ML) classifier is presented. The proposed technique allows the classifier's parameters, obtained by supervised learning on a specific image, to be upd...
详细信息
An unsupervised retraining technique for a maximum likelihood (ML) classifier is presented. The proposed technique allows the classifier's parameters, obtained by supervised learning on a specific image, to be updated in a totally unsupervised way on the basis of the distribution of a new image to be classified. This enables the classifier to provide a high accuracy for the new image even when the corresponding training set is not available.
With the discovery of single nucleotide polymorphisms (SNP) along the genome, genotyping of large samples of biallelic multilocus genetic phenotypes for (fine) mapping of disease genes or for population studies has be...
详细信息
With the discovery of single nucleotide polymorphisms (SNP) along the genome, genotyping of large samples of biallelic multilocus genetic phenotypes for (fine) mapping of disease genes or for population studies has become standard practice. A genetic trait, however, is mainly caused by an underlying defective haplotype, and populations are best characterized by their haplotype frequencies. Therefore, it is essential to infer from the phase unknown genetic phenotypes in a sample drawn from a population the haplotype frequencies in the population and the underlying haplotype pairs in the sample in order to find disease predisposing genes by some association or haplotype sharing algorithm. Haplotype frequencies and haplotype pairs are estimated via a maximum likelihood approach by a well-known expectationmaximization (EM) algorithm, adapting it to a large number (up to 30) of biallelic loci (SNP), and including nuclear family information, if available, into the analysis. Parents are treated as an independent sample from the population. Their genotyped offspring reduces the number of potential haplotype pairs for both parents, resulting in a higher accuracy of the estimation, and may also reduce computation time. In a series of simulations our approach of including nuclear family information has been tested against both the EM algorithm without nuclear family information and an alternative approach using GENEHUNTER for the haplotyping of the families, using the locus-by-locus allele counts of the sample. Our new approach is more precise in haplotyping in cases of a high number of heterozygous loci, whereas for a moderate number of heterozygous positions in the sample all three different approaches gave the same perfect results. Hum Mutat 17:289-295, 2001. (C) zool Wiley-Liss, Inc.
We consider a parametric family of multivariate density functions formed by mixture models from univariate functions of the type exp(-\x\(alpha)) for modeling acoustic feature vectors used in automatic recognition of ...
详细信息
We consider a parametric family of multivariate density functions formed by mixture models from univariate functions of the type exp(-\x\(alpha)) for modeling acoustic feature vectors used in automatic recognition of speech. The parameter alpha is used to measure the non-Gaussian nature of the data. Previous work has focused on estimating the mean and the variance of the data for a fixed alpha. Here we attempt to estimate the alpha from the data using a maximum likelihood criterion. Among other things, we show that there is a balance between a and the number of data points N that must be satisfied for efficient estimation. Numerical experiments are performed on multidimensional vectors obtained from speech data.
暂无评论