Brown University -Division of Biology and Medicine
Center for Statistical Sciences
*Joint Seminar with the Department of Economics
Massachusetts Institute of Technology | |
Reception following seminar at 167 Angell St., 2nd floor conference room |
Abstract: This paper develops a new instrumental variables method for treatment effects on quantiles and uses it to estimate effects of subsidized training on earnings distributions. The Quantile treatment effects (QTE) estimator reduces to quantile regression when treatment selection is exogenous. QTE solves a convex linear programming problem, after first step estimation of a nuisance function. We develop distribution theory assuming the first step is estimated nonparametrically. The empirical results for women show that training had the largest proportional impact at low quantiles. Perhaps surprisingly, however, training raised earnings quantiles for men only in the upper half of the earnings distribution.
*Abadie, Alberto; Angrist, Joshua D.; and Imbens, Guido W.
Stochastic Systems Seminar
Brown Applied Mathematics Pattern Theory and Vision Seminar
Center for Neural Science, and Courant Institute for Mathematical Sciences, New York University | |
Their Implications for Neural Representation | |
Abstract: I examine the statistics of natural monochromatic images decomposed using a multi-scale oriented basis. Although the coefficients of this representation are nearly decorrelated, they exhibit important higher-order statistical dependencies that cannot be eliminated with purely linear processing. In particular rectified coefficients corresponding to basis functions at neighboring spatial positions, orientations and scales are highly correlated. A method of removing these dependencies is to {\it divide} each coefficient by a weighted combination of its rectified neighbors. Several successful models of the steady-state behavior of neurons in primary visual cortex are based on such "divisive normalization" computations, and thus our analysis provides a theoretical justification for these models. Perhaps more importantly, the statistical measurements explicitly specify the weights that should be used in computing the normalization signal. I'll demonstrate that this weighting is qualitatively consistent with physiological experiments that characterize the suppressive effect of stimuli presented both within and beyond the classical receptive field. Similar relationships may be found in the auditory system. This framework provides an opportunity to directly test (through physiological predictions and comparisons) the ecological hypothesis that neural computations are optimally matched to the statistics of the environment.
Brown Analysis Seminar
LEMS and Electrical Sciences Seminar
Washington University, Saint Louis, Missouri | |
**NEW LISTING** |
Abstract: The problem of estimating the structure of objects moving relative to a camera and a laser range finder, with applications in mobile robotics and aerospace, has been a subject of great interest for well over two decades. To study this class of estimation problem, the role of a homogeneouse dynamical system will be emphasized, leading to a criterion which determines to what extent parameters can be identified. Dynamics of feature points, lines and algebraic curves in space, all have this flavor with varying degrees of complexity. This talk will emphasize the role of dynamical systems theory and algebraic curves in mobile robotics and machine vision.
Brown University Graduate School Dissertation Defense
Scientific Computing Seminar
Abstract: Mode-locked lasers are under intense scrutiny for use in communications systems. In theory, they should be capable of extremely high frequency, steady pulses which would be ideal for use in digital communications systems. A number of groups have built and examined these laser systems. There have been a growing number of reports on numerical approximations of the governing equations but few details on the numerics have been shared.
We will focus on the numerical approximation of the governing equations. The basic approaches to the discretization will be examined as well as the results of some of our numerical trials. In particular we will focus on how to implement a high-order method that makes use of Hermite polynomials and comparisons with a finite difference method will be given.
We will also discuss some of the difficulties we encountered in developing the numerical trials. The numerical trials are based on providing a single pulse and following its evolution in time. Comparisons to the numerical trials and approximations for parameterizations of the single-pulse dynamics will be examined.
*John B. Geddes, Kelly Black, and Matthew A. Beauregard
Department of Mathematics Undergraduate Colloquium
Department of Mathematics Colloquium
<--- 2001 Index