Адрес e-mail:

Lecture course "High-dimensional statistics"

Sara VAN DE GEER (Seminar for Statistics, ETH Zurich)

Dates: 22, 24, 26April 2013

Venue: Institute of Information Transmission Problems of Moscow, BolshojKaretnyi per.19.


09.00 – 10.00 Introduction: the high-dimensional linear model. Sparsity Oracle inequalities for the `1-penalized least squares estimator (Lasso) Coffee break
10.30 – 11.30 Variable selection and testing

24th April
09.00 – 10.00 High-dimensional non-linear models Tools from probability theory
Coffee break
10.30 – 11.30 Structured sparsity

26th April
09.00 – 10.00 Graphical models and causality
Coffee beak
10.30 - 1130 Sparse directed acyclic graphs

P. B?hlmannand S. van de Geer (2011). Statistics for High-Dimensional Data:
Methods, Theory and Applications. Springer
S. van de Geer and P. M?ller (2012). Quasi-likelihood and/or robust estimation
in high-dimensions. Statistical Science 27, 469-480
S. van de Geer and J. Lederer (2012). The Lasso, correlated design, and improved
oracle inequalities. IMS Collections From Probability to Statistics and
Back: High-Dimensional Models and Processes 9, 303-316
S. van de Geer, and J. Lederer (2013). The Bernstein-Orlicz norm and deviation
inequalities. To appear in Probability Theory and Related Fields.arXiv:1111.2450
S. van de Geer (2013). Generic chaining and the l1-penalty. To appear in Jour-
nal of Statistical Planning and Inference (discussion paper). arXiv:1205.3703
S. van de Geer (2013). Weakly decomposable regularization penalties and structured
sparsity. arXiv:1204.4813
S. van de Geer and P. B?hlmann (2013). l0-penalized maximum likelihood
for sparse directed acyclic graphs. To appear in The Annals of Statistics.
Если вы заметили в тексте ошибку, выделите её и нажмите Ctrl+Enter.

МФТИ в социальных сетях

soc-vk soc-fb soc-tw soc-li soc-li soc-yt