0.25
0.5
0.75
1.25
1.5
1.75
2
Gradient Weights help Nonparametric Regressors
Published on Jan 16, 20133344 Views
In regression problems over real d, the unknown function f often varies more in some coordinates than in others. We show that weighting each coordinate i with the estimated norm of the ith derivative
Related categories
Chapter list
Gradient weights help nonparametric regression00:00
Preliminaries (1)00:49
Preliminaries (2)01:01
Preliminaries (3)01:42
Motivation behind the approach (1)01:50
Motivation behind the approach (2)02:18
Motivation behind the approach (3)02:27
Motivation behind the approach (4)02:34
Motivation behind the approach (5)02:35
Motivation behind the approach (6)03:08
Gradient weighting (1)03:18
Gradient weighting (2)03:22
Gradient weighting (3)03:32
Gradient weighting (4)03:50
Why it works: local regression on (X; p). (1)04:18
Why it works: local regression on (X;p ). (2)04:44
Why it works: local regression on (X;p ). (3)04:57
Why it works: local regression on (X;p ). (4)05:23
Efficient estimation (1)05:45
Efficient estimation (2)06:04
Efficient estimation (3)06:54
Efficient estimation (4)07:14
Efficient estimation (5)07:49
Wi consistently estimates (1)08:07
Wi consistently estimates (2)08:56
Wi consistently estimates (4)09:04
Significant performance improvement in practice ... (1)09:21
Significant performance improvement in practice ... (2)09:37
On real-world data09:53
Results on real world datasets, kernel regression.10:29
Results on real world datasets, k-NN regression.11:27
MSE, varying training size, kernel regression12:11
MSE, varying training size, k-NN regression12:38
Take home message (1)13:05
Take home message (2)13:16
Take home message (3)13:26