Table of contentsText Learning
1. CoTraining learning from labeled and unlabeled data
Redundantly Sufficient Features
Slide 4
Slide 5
Slide 6
CoTraining Setting
Co-Training Rote Learner
Slide 9
Slide 10
Slide 11
Slide 12
What if CoTraining Assumption Not Perfectly Satisfied?
Slide 14
Slide 15
What Objective Function?
Slide 17
Slide 18
What Function Approximators?
Slide 20
PowerPoint Presentation
Classifying Jobs for FlipDog
Gradient CoTraining Classifying FlipDog job descriptions: SysAdmin vs. WebProgrammer
Gradient CoTraining Classifying Upper Case sequences as Person Names
CoTraining Summary
The Problem/Opportunity
Learn Local Regularities for Page Classification
Slide 28
Slide 29
Slide 30
Slide 31
Slide 32
Slide 33
Results of Local Learning: Cont.Education Course Page
Learning Site-Specific Regularities: Example 2
Slide 36
Local/Global Learning Algorithm
Slide 38
Slide 39
Slide 40
Slide 41
Local/Global Learning Summary
Other Local Learning Approaches
When/Why does this work??
Other Uses?
Summary
|