Workshop on Structure Adapting Methods - Abstract
High-dimensional data where the number of covariates is much larger than sample size occurs in many applications nowadays, for example in imaging, economics or biology. Remarkable progress has been achieved over the last few years in terms of statistical methodology, computation and theory. These developments had a substantial influence on high-dimensional data analysis in practice. Still, the statistical approach for extracting useful information from high-dimensional data is far from being complete. Major problems which often arise include: (i) a potential lack of stability of a solution; and (ii) specification of uncertainty and significance. After an introductory part describing achievements and limitations, the talk will address the issues (i) and (ii). We will present new methodology and theory, and we will apply these principles to problems from molecular biology.