Thesis (Ph. D.)--University of Chicago, Divinity School, 2001. / Includes bibliographical references. Also available on the Internet.

Thesis (Ph. D.)--University of California, San Diego, 2003. / Vita. Includes bibliographical references (leaves 106-109).

Thesis (Ph. D.)--York University, 2001. Graduate Programme in Statistics. / Typescript. Includes bibliographical references (leaves 96-103). Also available on the Internet. MODE OF ACCESS via web browser by entering the following URL: http://wwwlib.umi.com/cr/yorku/fullcit?pNQ66371.

In this dissertation, we study minimax hypothesis testing in high-dimensional regression against sparse alternatives and minimax estimation of average treatment effect in an semiparametric regression with possibly large number of covariates.

published_or_final_version / abstract / toc / Computer Science and Information Systems / Master / Master of Philosophy

Tikrinamos statistinės hipotezės. Darbas susideda iš dviejų dalių: pirmoje dalyje tikrinamos dvi paprastos hipotezės, taikomas statistinis kriterijus, kuris turi tenkinti tokias sąlygas: 1 – pirmos rūšies klaidos tikimybė yra aprėžta, o 2 - antros rūšies klaidos tikimybę mes minimizuojame. Darbe gauta asimptotinė formulė šios antros rūšies klaidos tikimybės elgesio, kai duomenų skaičius n artėja į begalybę. Antroje dalyje tikrinamos trys paprastos hipotezės ir ištirta kaip elgiasi minimali klaidos tikimybė iš visų galimų, kitaip sakant optimalaus minimaksiško kriterijaus klaidos tikimybė, kai duomenų skaičius n neapibrėžtai didėja. Pirmoje dalyje nagrinėjama klaidos tikimybė asimtotiškai mažėja proporcingai Kulbako atstumui, o antroje dalyje atitinkama klaidos tikimybė atitinkamai mažėja proporcingai Čensovo atstumui. Taip pat savo darbo teorinėje dalyje aprašiau šių atstumų savybes. Pabaigoje, kaip pavyzdį paėmus polinominį skirstinį parodėme, kaip yra surandamos šių dviejų rezultatų asimptotinės formulės ir kaip atrodo atitinkamų šių hipotezių tikrinimo asimptotinis elgesys. / Statistical Hypothesis Testing. The work consists of two parts: in the first part two simple hypotheses were tested and a statistical criterion which should meet the following specifications was applied: 1 – the probability of type I error was defined and 2 - the probability of type II error was minimized. The asymptote formula of behaviour of the probability of type II error was obtained in the work when the n number of the data was approaching infinity. In the second part of the work three simple hypotheses were tested. It has been explored how the minimum error probability of all available errors, i.e., the error probability of the optimal minimax criterion, performed when the n number of the data was indefinitely increasing. The error probability analysed in the first part of the work was asymptotically decreasing in proportion to the Kulbak distance while the corresponding error probability analysed in the second part of the work was respectively decreasing in proportion to the Censov distance. Moreover, the properties of the distances mentioned above were described in the theoretical part of the work. The work concludes with choosing a polynomial distribution as an example which has demonstrated how the two asymptote formulas of the two results were obtained and how the asymptotic behaviour of testing of respective hypotheses looked like.

Prompted by oculomotor theories of attention, the present experiments explore the role of saccade activation in the generation of two cueing effects: exogenous capture (Experiment 1) and inhibition of return (IOR; Experiment 2). Exogenous capture is shortlived and marked by faster responding toward recently stimulated locations, whereas the longer-lasting IOR manifests as slower responding toward those locations. Within each experiment, Group A performed in a dual-task in which on most trials a peripheral target had to be identified but infrequently a central arrow probe called for an eye movement instead, while for Group B the tasks were the same except saccade trials were frequent and target identification trials were infrequent. In Experiment 1, for group A uninformative cues captured attention as measured by faster digit identification at the cued location, an effect not accompanied by saccade activation. For group B, cues generated saccade activation without capturing attention. Thus saccade activation need not accompany exogenous covert capture, and covert capture need not accompany saccade activation. In Experiment 2, group A exhibited IOR which slowed digit identification, but did not affect saccadic responding, while Group B exhibited no IOR in either digit identification or eye movement trials. This finding provides converging evidence that IOR can be dichotomized into two forms; one which delays motor production itself (Evidenced amply elsewhere, e.g., Taylor & Klein, 2000) and another which delays responding by applying inhibition at a perceptual-motor interface which can operate in independence from its motoric cousin.

Mud-pulse telemetry is a method used for measurement-while-drilling (MWD)in the oil industry. The telemetry signals are corrupted by spurious mud pump noise consisting of a large number of harmonically related sinusoids. In order to denoise the signal, the noise parameters have to be tracked accurately in real time. There are well established parametric estimation techniques for determining various parameters of independent sinusoids. The iterative methods based on the linear prediction properties of the sinusoids provide a computationally e±cient way of solving the non linear optimization problem presented by these methods. However, owing to the large number of these sinusoids, incorporating the harmonic relationship in the problem becomes important. This thesis is aimed at solving the problem of estimating parameters of harmonically related sinusoids. We examine the efficacy of IQML algorithm in estimating the parameters of the telemetry signal for varying SNRs and data lengths. The IQML algorithm proves quite robust and successfully tracks both stationary and slowly varying frequency signals. Later, we propose an algorithm for fundamental frequency estimation which relies on the initial harmonic frequency estimate. The results of tests performed on synthetic data that imitates real field data are presented. The analysis of the simulation results shows that the proposed method manages to remove noise causing sinusoids in the telemetry signal to a great extent. The low computational complexity of the algorithm also makes for an easy implementation on field where computational power is limited.

ABSTRACT Introduction: Human newborn and animal studies provide support for the intersensory redundancy hypothesis, which posits that learning is more effective when information is presented simultaneously in two modalities than one alone. Whether the same is true in the human fetus is unknown and was examined in this study. Methods: 63 low-risk fetuses (≥36 weeks gestation) were randomly assigned to one of 6 experimental groups: each group included one of 3 stimulus conditions [unimodal (music), unimodal (maternal sway) or bimodal (music and maternal sway)], and one of 2 pieces of music (music A, 4/4 time; music B, 3/4 time) composed for the study. Laboratory pre-testing included a 2 min no-music, 2 min music (A or B), 2 min no-music observation while fetal heart rate (FHR) and body movements were recorded. Subsequently, mothers carried out the assigned intervention at home, twice a day for 5 days. On day 6, laboratory testing was repeated first with the familiar (A or B) and then the novel music. Results: The initial testing showed a difference between Music A and Music B, F (1, 61) = 8.203, p <.01, where FHR decreased to Music A and increased to Music B. The same FHR response was found when fetuses were exposed to the opposite music for the first time in the novelty testing, F (1, 44) = 4.543, p <.05, following intervention. Music A elicited a response in both the unimodal music only and sway only groups, F (29, 203) = 1.871, p < .01, and F (29, 174) = 1.818, p < .01, respectively. In music B only the multimodal group showed an effect of intervention, F = (29, 203) = 1.914, p < .005. Conclusions: Fetal response to music A and B was qualitatively different. During pretesting, FHR decreased to music A and increased to music B. When the stimulus elicited an attention response (FHR decrease) learning was observed in both the unimodal or multimodal conditions. This is seen with music A (4/4 time) music where the fetus learns the stimulus. When the stimulus did not elicit a FHR decrease (Music B, 3/4 time), there was evidence that a multimodal stimulus was more effective providing some support for the intersensory redundancy hypothesis. / Thesis (Master, Nursing) -- Queen's University, 2007-10-18 16:52:55.68

No description available.