Global ETD Search

1	Variable selection in two-group discriminant analysis using the linear discriminant function Ganeshanandam, S. January 1987 (has links) No description available. 519.5 Error rate estimation
2	On Applications of Semiparametric Methods Li, Zhijian 01 October 2018 (has links) No description available. Statistics Semiparametric error variance estimation nonparametric dimension reduction
3	Nonlinear Bounded-Error Target State Estimation Using Redundant States Covello, James Anthony January 2006 (has links) When the primary measurement sensor is passive in nature--by which we mean that it does not directly measure range or range rate--there are well-documented challenges for target state estimation. Most estimation schemes rely on variations of the Extended Kalman Filter (EKF), which, in certain situations, suffer from divergence and/or covariance collapse. For this and other reasons, we believe that the Kalman filter is fundamentally ill-suited to the problems that are inherent in target state estimation using passive sensors. As an alternative, we propose a bounded-error (or set-membership) approach to the target state estimation problem. Such estimators are nearly as old as the Kalman filter, but have enjoyed much less attention. In this study we develop a practical estimator that bounds the target states, and apply it to the two-dimensional case of a submarine tracking a surface vessel, which is commonly referred to as Target Motion Analysis (TMA). The estimator is robust in the sense that the true target state does not escape the determined bounds; and the estimator is not unduly pessimistic in the sense that the bounds are not wider than the situation dictates. The estimator is--as is the problem itself--nonlinear and geometric in nature. In part, the simplicity of the estimator is maintained by using redundant states to parameterize the target's velocity. These redundant states also simplify the incorporation of other measurements that are frequently available to the system. The estimator's performance is assessed in a series of simulations and the results are analyzed. Extensions of the algorithm are considered. target state estimation target motion analysis bearings-only ranging bounded-error state estimation
4	The Consistency ot the Empirical Quantization Error Pötzelberger, Klaus January 1999 (has links) (PDF) We study the empirical quantization error in case the number of prototypes increases with the size of the sample. We present a proof of the consistency of the empirical quantization error and of corresponding estimators of the quantization dimensions of distributions. (author's abstract) / Series: Forschungsberichte / Institut für Statistik MSC 60F15, 60F05, 62G20, 28A75
5	保險法中股票投資限制與估計風險之探討郭榮堅, Kuo, Jung-Ching Unknown Date (has links) 假設真實輸入參數可以事先預知，實際上，透過平均數-變異數的投資組合量化模式，效率前緣將會往右下方移動。這也意謂倘若保險公司的投資策略有達到效率前緣的情況下，保險法146條的投資上限將使得保險公司的投資報酬率降低，以及投資風險增加。給定投資上限反而降低保險公司的投資績效。然而真實輸入參數並無法事先預知，因此根據過去的經驗資料以及主觀判斷來估計輸入參數將是作為取代真實參數的作法。而估計誤差的存在將勢難避免。同時平均數-變異數模式所決定的投資組合會過度投資在高估投資報酬率以及低估投資風險的股票上，因此估計誤差的影響將是不容忽略。並且保險公司在追求資產極大化的同時，有其必要兼顧到估計誤差的影響。本研究主要有三個主題，首先我們透過平均數-變異數模式來探討估計誤差對投資組合的影響。緊接著就投資上限與估計風險的關係進行研究。最後分析股票淨利率限制和投資上限之間的關係，並且解釋存在的功能以及探討是否有其存在的必要性。本研究結果發現現行的單一股票的投資上限(七分之一)不但降低了投資風險，並且改善了真實反應的投資報酬率低於6%，甚至為負值的問題。並且就淨利率上限為6%的合適性而言，除非保險公司採取的投資策略非常的保守，否則還不如放寬，或者廢除。投資限制保險法估計誤差效率前緣股票 limit of investment the law of insurance error of estimation efficient fronter stock estimation risk
6	NOVÉ METODY KANÁLOVÉHO KÓDOVÁNÍ PRO DRUŽICOVOU KOMUNIKACI / NEW CHANNEL CODING METHODS FOR SATELLITE COMMUNICATION Rumánek, Jaroslav January 2010 (has links) This dissertation thesis deals with new progressive channel coding methods for data transmission using satellite transponder. The design of the system for SMS transmission, in which novel turbo coding methods are applied, is discussed too. An achievement of the lowest output power and the smallest user aperture is the principal aim of the new method applications. Design of system that would be able to the error free SMS transmission by very low signal to noise ratio is analyzed in this dissertation thesis. The work is focused on energy budget, modification and implementation of new turbo code types and using unique properties, development of new bit error rate estimation methods and methods for determination of final SMS form. The main contribution is the new type of turbo code development that have optimal properties for this usage, development of new bit error rate estimation method and development of method that is able to determine final form of SMS on the basis SMS frame structure and turbo decoding theory when the bit error rate is not zero.
7	Statistical Inference Chou, Pei-Hsin 26 June 2008 (has links) In this paper, we will investigate the important properties of three major parts of statistical inference: point estimation, interval estimation and hypothesis testing. For point estimation, we consider the two methods of finding estimators: moment estimators and maximum likelihood estimators, and three methods of evaluating estimators: mean squared error, best unbiased estimators and sufficiency and unbiasedness. For interval estimation, we consider the the general confidence interval, confidence interval in one sample, confidence interval in two samples, sample sizes and finite population correction factors. In hypothesis testing, we consider the theory of testing of hypotheses, testing in one sample, testing in two samples, and the three methods of finding tests: uniformly most powerful test, likelihood ratio test and goodness of fit test. Many examples are used to illustrate their applications. finite population correction factor statistical hypothesis type I error composite hypothesis power Central limit theorem Basu theorem critical value goodness of fit test likelihood ratio test Lehmann-Scheffe theorem complete statistic ancillary statistic P-value type II error Rao-Blackwell theorem simple hypothesis one-sided test two-sided test significance level sufficient statistic alternative hypothesis testing hypothesis null hypothesis rejection region confidence interval confidence level minimum variance unbiased estimator interval estimation uniformly most powerful test Cramer-Rao inequality likelihood function mean square error moment estimator error of estimation point estimation maximum likelihood estimator bias consistent estimator minimal sufficient statistic factorization theorem unbiased estimator statistic most powerful test test statistics Neyman-Pearson lemma decision rule

1

Page generated in 0.1467 seconds