Deviation Optimal Learning using Greedy Q-aggregation

Details Deviation Optimal Learning using Greedy Q-aggregation Deviation Optimal Learning using Greedy Q-aggregation

2012-03-12

arxiv.org/abs/1203.2507v1

Mathematics

Statistics Theory

Scientific paper

Given a finite family of functions, the goal of model selection is to construct a procedure that mimics the function from this family that is the closest to an unknown regression function. More precisely, we consider a general regression model with fixed design and measure the distance between functions by the mean squared error at the design points. While procedures based on exponential weights are known to solve the problem of model selection in expectation, they are, surprisingly, sub-optimal in deviation. We propose a new formulation called Q-aggregation that addresses this limitation; namely, its solution leads to sharp oracle inequalities that are optimal in a minimax sense. Moreover, based on the new formulation, we design greedy Q-aggregation procedures that produce sparse aggregation models achieving the optimal rate. The convergence and performance of these greedy procedures are illustrated and compared with other standard methods on simulated examples.

Affiliated with

Also associated with

No associations

Rating

Deviation Optimal Learning using Greedy Q-aggregation does not yet have a rating. At this time, there are no reviews or comments for this scientific paper.

If you have personal experience with Deviation Optimal Learning using Greedy Q-aggregation, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Deviation Optimal Learning using Greedy Q-aggregation will most certainly appreciate the feedback.

Rate now

Comments { 0 }

Profile ID: LFWR-SCP-O-488472