0.25
0.5
0.75
1.25
1.5
1.75
2
Optimal aggregation of affine estimators
Published on Aug 02, 20113019 Views
We consider the problem of combining a (possibly uncountably infinite) set of affine estimators in non-parametric regression model with heteroscedastic Gaussian noise. Focusing on the exponentially w
Related categories
Chapter list
Optimal Aggregation of Affine Estimators00:00
Introduction00:01
Motivations : doing as good as the best filter00:54
Motivations : doing as good as the best dictionary approximation01:42
Penalization Methods02:26
Mixing classical filtering and dictionary learning : known results04:25
Notation and model05:25
Aggregation of Estimators and Oracle Inequalities06:34
EWA : classical point of view08:48
EWA : Penalty point of view - 110:36
EWA : Penalty point of view - 211:18
Affine estimators11:48
Linear case - 112:36
Linear case - 213:02
Main theorem conditions - 113:26
Main theorem conditions - 213:38
Main Theorem14:01
Corollary : finite case15:02
Minimax point of view16:00
Corollary : Adaptation17:14
1D signal experiments18:45
Conclusion - 119:27
Conclusion - 220:07