|Date/Time:||Monday, 06 Apr 2015 from 4:10 pm to 5:00 pm|
|Actions:||Download iCal/vCal | Email Reminder|
For high dimensional regression, the identification of interaction effects is extremely challenging in terms of both computation and theoretical investigation. We study the key issues and reveal some interesting results. Then we propose new methods for high-dimensional interaction selection, which are feasible even when the dimension increases exponentially fast with the sample size. This talk demonstrates several features of these methods: fast computation, rigorous theoretical support, and competitive numerical performances.