Brown University Center for Statistical Sciences Seminar
Carnegie Mellon University, Pittsburgh | |
1st Floor Conference Room *Sponsored by the Charles K. Colver Lecturship Fund |
Abstract: Multiple testing problems arise frequently in modern applications. A common analysis of functional neuroimaging data involves performing a hypothesis test at every volume element in the brain to locate brain activity. Radio astronomers perform tests at every location in a map to distinguish sources from background. In the analysis of DNA microarrays, comparisons across experimental groups involve simultaneous tests at thousands of genes.
The challenge of multiple testing is to define a decision rule that provides good power while controlling some overall measure of error. Traditional methods seek strong control of the familywise error rate. Benjamini & Hochberg (1995) introduced a new criterion -- the False Discovery Rate -- and put forward a procedure to control it. The False Discovery Rate is the expected proportion of false discoveries among the rejected null hypotheses. Since then, false discovery control has become an active area of research.
I will introduce methods for controlling other features of the false discovery distribution. A key to the approach is to study the proportion of false discoveries as a stochastic process in the rejection threshold. I will describe methods for constructing confidence envelopes for this unobserved process and techniques for constructing thresholds with desirable properties. The techniques extend in an interesting way to spatial and clustering problems on random fields.
*This is joint work with Larry Wasserman at Carnegie Mellon Statistics.
Brown University-
Joint Materials/Solid Mechanics Seminar Series
Abstract: A three-dimensional crystal-mechanics based theory for the thermo-mechanically coupled superelastic response of polycrystalline shape-memory materials is developed and used to simulate the response of a Ti-Ni (Nitinol) shape -memory alloy. Both manifestations of superelasticity: stress-strain response at a fixed temperature, and strain- temperature response at a fixed stress have been experimentally studied. The model, when suitably numerically implemented and calibrated, is shown to accurately predict the superelastic response of the material. Also, the strain-temperature cycling experiments under different constant axial stresses are predicted with reasonable accuracy. The effects of self-heating and cooling due to the exothermic and endothermic nature of the austenite-to-martensite and martensite-to-austenite transformations were investigated by performing superelastic tension experiments at strain rates which are high enough to result in non-isothermal testing conditions. The thermo- mechanically coupled theory is able to capture the resulting inhomogeneous deformation associated with the nucleation and propagation of transformation fronts, and also the "apparent hardening" of the nominal stress-strain curves observed in the experiments.
Neural Coding/Pop Code Meeting
<--- 2003 Index