Tuesday, May 29, 2018, 1:00 pm — Training Room, Building 725
We describe a procedure enhancing L1-penalized regression by adding dynamically generated rules describing multidimensional "box" sets. In contrast to prior approaches to this class of problems, we draw heavily on standard mathematical programming techniques, enhanced by parallel computing. Our rule-adding procedure is based on the classical column generation method for high-dimensional linear programming. The pricing problem for our column generation procedure reduces to the NP-hard rectangular maximum agreement (RMA) problem of finding a box that best discriminates between two weighted datasets. We solve this problem exactly using a parallel branch-and-bound procedure or approximately by a greedy heuristic. The resulting rule-enhanced regression method is computation-intensive, but our computational tests suggest that outperforms prior methods at making accurate and stable predictions.
Hosted by: Kerstin Kleese van Dam
14131 | INT/EXT | Events Calendar
Not all computers/devices will add this event to your calendar automatically.
A calendar event file named "calendar.ics" will be placed in your downloads location. Depending on how your device/computer is configured, you may have to locate this file and double click on it to add the event to your calendar.
Event dates, times, and locations are subject to change. Event details will not be updated automatically once you add this event to your own calendar. Check the Lab's Events Calendar to ensure that you have the latest event information.