Peter Bickel
statisticshigh-dimensionalsemiparametricgenomicsberkeley
Abstraction: UC Berkeley statistician's research spanning semiparametrics, high-dimensional inference, genomics
Key points:
- Co-author of the book "Efficient and Adaptive Estimation for Semiparametric Models"; uses asymptotic theory to guide model development
- Major contributions to high-dimensional covariance estimation: regularization by thresholding (2008, Annals of Statistics) and banding/regularization of large covariance matrices
- Simultaneous analysis of Lasso and Dantzig Selector (2009, Annals of Statistics, with Ritov and Tsybakov)
- Work on curse of dimensionality in particle filters: collapse of importance sampling in very large scale systems
- Nonparametric network models and Newman-Girvan modularity (PNAS 2009, with Chen)
- Applied genomics: ENCODE pilot project (Nature 2007), nonparametric methods for genomic inference
Connections: Peter Bickel · Uc Berkeley · High Dimensional Statistics · Semiparametric Models · Covariance Estimation