0.25
0.5
0.75
1.25
1.5
1.75
2
Stumping along a summary
Published on Jul 25, 20113109 Views
The methods we used to compete in the « Exploration & Exploitation » challenge are based on three layers. The first layer provides an online summary of the data stream for continuous and nominal data.
Related categories
Chapter list
Stumping Along a Summary Exploration & Exploitation Challenge 201100:00
Outline00:17
Online content selection problem01:14
Off-line evaluation of online content selection systems01:19
Outline: Approach (1)01:20
Motivation for a Summary01:28
Numerical features summary01:53
Nominal features summary03:22
Outline: Approach (2)04:13
Base Predictors: probability estimation trees04:20
Outline: Approach (3)06:00
Online combination of predictors06:15
Experiments07:27
Outline: Experiments and Results (1)08:11
Adobe dataset with perfect player08:14
Adobe dataset without perfect player08:54
Deviation from static model09:49
Utree vs. full dataset discretization (MODL)10:28
Outline: Experiments and Results (2)11:21
Orange portal dataset with perfect player11:25
Orange dataset with perfect player11:53
Conclusion and discussion12:05