465 Jon M. Huntsman Hall
3730 Walnut Street
Philadelphia, PA 19104
Research Interests: Statistical methods for big data, High-dimensional asymptotics, Random matrix theory, Multiple testing
A postdoctoral research fellow position is available. The candidate will work on developing statistical methods and theory for massive “big” datasets. The position is for a two year period starting Summer 2019. The candidate will have the opportunity to collaborate with other faculty in the Statistics department, and collaborators across the University of Pennsylvania. Desired qualifications include a PhD in Statistics, Applied Math, Electrical Engineering, or related fields, and a promising publication record. If interested, please send CV, cover letter, and two letters of reference to Edgar Dobriban. See here for a more formal announcement: https://forms.stat.ufl.edu/statistics-jobs/entry/5534/
I am actively looking for motivated students. If you are already at Penn, please feel free to contact me about possible research projects.
Current (co-)advisees: Yue Sheng (AMCS), Sifan Liu (Tsinghua undergraduate).
New seminar class in Fall 2018 and Spring 2019: Topics in Deep Learning (STAT-991), surveying advanced topics in deep learning research based on student presentations. See here for the syllabus and for the lecture notes.
Research projects for exceptional undergraduates are available.
I am also looking for collaborators, both in statistics and in other areas. Please feel free to contact me about potential problems of mutual interest.
Talk slides: GitHub.
Edgar Dobriban and Stefan Wager (2018), High-dimensional asymptotics of prediction: ridge regression and classification, The Annals of Statistics, 46 (1), pp. 247-279.
Edgar Dobriban (2017), Weighted mining of massive collections of p-values by convex optimization, Information and Inference: A Journal of the IMA, 7 (2), pp. 251-275.
Edgar Dobriban and Art B. Owen (2017), Deterministic parallel analysis: An improved method for selecting the number of factors and principal components, Journal of the Royal Statistical Society - Series B (JRSS-B).
Edgar Dobriban, William Leeb, Amit Singer (2017), Optimal prediction in the linearly transformed spiked model, The Annals of Statistics, to Appear.
Description: This paper supersedes the older Dobriban, Leeb, Singer manuscript "PCA from noisy, linearly reduced data: the diagonal case".
This seminar will be taken by doctoral candidates after the completion of most of their coursework. Topics vary from year to year and are chosen from advance probability, statistical inference, robust methods, and decision theory with principal emphasis on applications.
This page has links to methods from my papers. Feel free to contact me if you are interested to use them.
The ePCA method for principal component analysis of exponential family data, e.g. Poisson-modeled count data. (with L.T. Liu);
Methods for working with large random data matrices, including
P-value weighting techniques for multiple hypothesis testing. These can improve power in multiple testing, if there is prior information about the individual effect sizes. Includes the iGWAS method for Genome-Wide Association Studies.