Browsing by Author "Lo, Wen-Hui"

Now showing 1 - 5 of 5

Advanced Metering the Signal Activity of Combined Signal in Sparse Data Condition
(2011)
Lo, Wen-Hui
;
Chen, Sin-Horng
;
電信工程研究所
;
Institute of Communications Engineering
The performance of finding the upper bound of eigenvalues (UBE) is affected by the quality of correlation matrix estimation. In this paper, a new quantile-based maximum likelihood mean estimator is proposed to improve the mean estimation on sparse data condition so as to obtain a more reliable correlation matrix estimate from observed samples. This in turn improves the UBE finding. The study is specially focused on the quasi-normal signal of combined quantities with asymptotic window-shape distribution and fast-decaying short tail. Experimental results show that the new mean estimator outperforms the conventional sample mean estimator on mean estimation. The UBE finding is also improved accordingly.
Robust Estimation for Sparse Data
(2008)
Lo, Wen-Hui
;
Chen, Sin-Horng
;
交大名義發表
;
National Chiao Tung University
Robust parameters estimation of sparse data is generally applied to the test cases of time-consuming or high cost data collection. This study concerns with the problem in small sample size which is often encountered in the client data processing for speaker verification. We found that there always exists a coverage mismatch problem between the samples and its population in terms of probability density function (pdf) when the sample size is less than 20. We call this special problem the distribution mismatch (DM) problem. The paper proposes to solve the DM problem through addressing a new coverage-based estimator.
The Mean Estimation of the Combined Quantities by the Asymptotic Minimax Optimization
(2009)
Lo, Wen-Hui
;
Chen, Sin-Horng
;
交大名義發表
;
National Chiao Tung University
The mean value estimation for the output quantity of combined random variables is one of the major issues in measurement. In this paper, a new quantile-based maximum likelihood estimation (QMLE) method for mean value estimation is proposed. It fuses the concept of both empirical and symmetric quantile to incorporate the order statistics into the QMLE. Unlike Sample mean derived basing only on the maximum likelihood criterion, the QMLE also considers MMSE defined using the quasi symmetric quantiles (QSQ), i.e., the first- and last-order samples. Simulation results confirm that the proposed QMLE mean estimator outperforms the conventional Sample mean estimator. This work also gives a looking-up table for the refinement corresponding to the QSQ adjustments.
The wide-sense parametric coverage estimator against the distribution mismatch problem for sparse data
(2008)
Lo, Wen-Hui
;
Chen, Sin-Horng
;
電信工程研究所
;
Institute of Communications Engineering
Robust parameter estimation of sparse data is generally applied to the tasks when data collection is time-consuming or of high cost. We point out a new problem caused by sparse data. We find that there may exists coverage mismatch between data samples and the population when the sample size is less than 20. We call it the distribution mismatch (DM) problem. In this study, we derive a wide-sense joint pdf for coverage, range, the sample of minimum order, and data samples themselves to analyze the DM problem. Based on the formulation, a new algorithm is proposed to compensate the DM problem. Experimental results show that the mean estimate of the algorithm will converge to the population mean if the standard deviation of population is known.
Theoretical and Practical Realization for the Uncertainty Measurement by Coverage Interval
(2009)
Lo, Wen-Hui
;
Chen, Sin-Horng
;
電信工程研究所
;
Institute of Communications Engineering
Uncertainty measurement and expression can be considered in different ways, but by using coverage interval is the most popular method if considering the bounding information. In this paper, we derive the theoretical probability distribution function of coverage interval and verify that its accuracy meets the best of fit. We illustrate that it may help the decision of the curve shape for the general expanded uncertainty task. We also suggest a new algorithm for the purpose of connecting the coverage interval to the statistical coverage interval which is robust on the confidence level representation for sparse data. In final, we extend the statistical coverage interval to a new more accurate representation of probability propagation of coverage interval. Experiments showed this new approach was robust especial in the sparse data condition.