Brown University Center for Statistical Sciences Seminar
Department of Biostatistics, University of Michigan School of Public Health | |
Abstract: The lack of an agreed inferential basis for statistics makes life "interesting" for academic statisticians, but at the price of negative implications for the practice and status of statistics in industry, science and government. Our discipline will only mature when we come to a basic agreement about how to apply statistics to real problems. Some illustrations and implications of the existing Bayes/frequentist rift, specific and general, are offered in in areas with which the speaker is familiar, including survey inference and missing data. Strengths and weaknesses of the frequentist and Bayes paradigms will be outlined. Can the rift be breached? In the talk a roadmap for a possible frequentist/Bayes compromise is provided based on the work of Box, Rubin and others. The compromise is sometimes called "calibrated Bayes", and asserts that inferences should be Bayesian and model-based, but model formation and assessment can and should involve frequentist ideas. Some implications of this proposed compromise for the future teaching and practice of statistics are offered.
Sponsored by the Charles K. Colson Lectureship and Publication Fund
Co-Sponsored by the Bruce M. Bigelow Class of 1955 Lecture Series
The Center for Computational Molecular Biology Seminar
Department of Electrical Engineering and Computer Science, Massachusetts Institute of Technology, Computer Science and Artificial Intelligence Laboratory | |
Abstract: We propose a game-theoretic approach to learn and predict coordinate binding of multiple DNA binding regulators. The framework implements resource constrained allocation of proteins to local neighborhoods as well as to sites themselves, and explicates coordinate and competitive binding relations among proteins with affinity to the site or region. Our model permits us to make numerical predictions genome-wide under different perturbations.
This talk will emphasize the mathematical and computational foundations of the new modeling approach. I will start by formally presenting our proposed model: the DNA Binding game. I will establish its ability to make predictions under any perturbations by showing that an equilibrium exists in any instantiation of the game. I will present in some detail a simple iterative algorithm that monotonically converges to an equilibrium of the game (thus providing a constructive proof of existence). Time permitting, I will show a small-scale illustration of our approach on a well-known biological subsystem: lambda-phage. I will conclude by briefly discussing work in progress on learning games from data to address large-scale biological problems.
Joint work with Luis Perez-Breva, Chen-Hsiang Yeang and Tommi Jaakkola.
Hosted by: Professor Charles Lawrence
<--- 2006 Index