On l1 mean and variance filtering
نویسندگان
چکیده
This paper addresses the problem of segmenting a time-series with respect to changes in the mean value or in the variance. The first case is when the time data is modeled as a sequence of independent and normal distributed random variables with unknown, possibly changing, mean value but fixed variance. The main assumption is that the mean value is piecewise constant in time, and the task is to estimate the change times and the mean values within the segments. The second case is when the mean value is constant, but the variance can change. The assumption is that the variance is piecewise constant in time, and we want to estimate change times and the variance values within the segments. To find solutions to these problems, we will study an l1 regularized maximum likelihood method, related to the fused lasso method and l1 trend filtering, where the parameters to be estimated are free to vary at each sample. To penalize variations in the estimated parameters, the l1-norm of the time difference of the parameters is used as a regularization term. This idea is closely related to total variation denoising. The main contribution is that a convex formulation of this variance estimation problem, where the parametrization is based on the inverse of the variance, can be formulated as a certain l1 mean estimation problem. This implies that results and methods for mean estimation can be applied to the challenging problem of variance segmentation/estimation. Copyright 1998 IEEE. Published in the Proceedings of the 45th Annual Asilomar Conference on Signals, Systems, and Computers, November 6-9, 2011Pacific Grove, California, USA Personal use of this material is permitted. However, permission to reprint/republish this material for advertising or promotional purposes or for creating new collective works for resale or redistribution to servers or lists, or to reuse any copyrighted component of this work in other works, must be obtained from the IEEE.
منابع مشابه
On-Line Nonlinear Dynamic Data Reconciliation Using Extended Kalman Filtering: Application to a Distillation Column and a CSTR
Extended Kalman Filtering (EKF) is a nonlinear dynamic data reconciliation (NDDR) method. One of its main advantages is its suitability for on-line applications. This paper presents an on-line NDDR method using EKF. It is implemented for two case studies, temperature measurements of a distillation column and concentration measurements of a CSTR. In each time step, random numbers with zero m...
متن کاملA Stock Market Filtering Model Based on Minimum Spanning Tree in Financial Networks
There have been several efforts in the literature to extract as much information as possible from the financial networks. Most of the research has been concerned about the hierarchical structures, clustering, topology and also the behavior of the market network; but not a notable work on the network filtration exists. This paper proposes a stock market filtering model using the correlation - ba...
متن کاملGlobal Constraints for the Mean Absolute Deviation and the Variance: Application to the Vertical Line Balancing
Optimization with a balancing objective often appear in practical problems where humans are implied in the solution. For example, in tasks assignment problems it is a desirable property that the workload is fairly distributed among the workers. In general, a violation measure of the perfect balance can be defined as the Lp norm of the vector of variables minus their mean. Two global constraints...
متن کاملL1-Norm Quantile Regression
Classical regression methods have focused mainly on estimating conditional mean functions. In recent years, however, quantile regression has emerged as a comprehensive approach to the statistical analysis of response models. In this article we consider the L1-norm (LASSO) regularized quantile regression (L1-norm QR), which uses the sum of the absolute values of the coefficients as the penalty. ...
متن کاملImproved feature enhancement using temporal filtering in speech recognition
The difference between training and testing environments is the major reason of performance degradation of speech recognition. In this paper, to further decrease the mismatch, we apply temporal filtering, Auto-Regression and Moving-Average (ARMA) filtering or RelAtive SpecTrAl (RASTA) filtering, as a post-processor for the log-Energy dynamic Range Normalization-Cepstral Mean and Variance Normal...
متن کامل