en-de
en-es
en-fr
en-pt
en-sl
en
en-zh
0.25
0.5
0.75
1.25
1.5
1.75
2
Oracle inequalities for computationally budgeted model selection
Published on Aug 02, 20113424 Views
We analyze general model selection procedures using penalized empirical loss minimization under computational constraints. While classical model selection approaches do not consider computational aspe
Related categories
Chapter list
Oracle inequalities for computationally budgeted model selection00:00
Model Selection - 100:00
Model Selection - 200:09
Model Selection - 300:13
Complexity penalized model selection01:01
Statistical dream, computational nightmare - 102:46
Statistical dream, computational nightmare - 204:04
Statistical dream, computational nightmare - 304:24
Statistical dream, computational nightmare - 404:43
A budgeted model selection framework - 104:56
A budgeted model selection framework - 205:41
A computational oracle inequality framework05:59
Naive solution: grid search - 107:18
Naive solution: grid search - 207:56
Model selection from nested classes08:42
Exploiting structure of nested classes09:30
Coarse grid sets10:51
Coarse-grid sets for model selection12:46
Oracle inequality for coarse grid search - 113:56
Oracle inequality for coarse grid search - 214:22
Corollary for logarithmic coarse grids - 115:27
Corollary for logarithmic coarse grids - 215:41
Removing the nesting assumption - 115:52
Removing the nesting assumption - 216:36
K-armed bandits for model selection17:02
Oracle inequality under separation assumption18:50
Oracle inequality without separation20:39
Conclusion22:49
Thank You23:50