en
0.25
0.5
0.75
1.25
1.5
1.75
2
Fast projections onto l1,q-norm balls for grouped feature selection
Published on Oct 03, 20113220 Views
Joint sparsity is widely acknowledged as a powerful structural cue for performing feature selection in setups where variables are expected to demonstrate “grouped” behavior. Such grouped behavior is
Related categories
Chapter list
Fast Projections onto l1, q-norm Balls00:00
Regularized Optimization (1)00:18
Regularized Optimization (2)00:43
Minimize (1)01:08
Minimize (2)01:50
We solve02:00
Main iteration (1)02:49
Main iteration (2)03:52
Main iteration (3)05:03
Mixed norms (1)05:48
Mixed norms (2)06:52
We need to solve07:52
Lagrangian duality08:14
Projections for mixed norms08:35
Finding 09:31
Root finding09:48
Theorem10:56
Numerical results (1)11:23
Numerical results (2)12:34
Numerical results (3)12:50
Multitask Lasso13:04
Synthetic Results (1)14:39
Synthetic Results (2)15:03
Synthetic Results (3)15:04
MTL on news data (1)15:12
MTL on news data (2)16:20
Summary17:08
The End!18:17