0.25
0.5
0.75
1.25
1.5
1.75
2
Learning with structured inputs
Published on Feb 25, 20073376 Views
I will present a novel approach to semi-supervised learning that employs a method which we refer to as structural learning (aka multi-task learning). The idea is to learn predictive structures from
Chapter list
Learning with Structured Input00:02
Outline00:57
Supervised learning01:47
Supervised learning01:54
Semi-supervised learning problem01:59
Chunking02:13
Previous semi-supervised approaches to NLP 03:45
Our approach04:49
Multi-task learning problem05:50
Multi-task learning (structural learning)06:09
Standard linear prediction model06:24
Example input vector representation07:40
New model for multi-task learning08:10
Joint empirical risk minimization10:10
Theoretical Justification11:16
Alternating Structure Optimization Algorithm (ASO)12:27
Semi-supervised learning13:13
Alternating Structure Optimization Algorithm (ASO)14:51
Semi-supervised learning19:00
Create multiple tasks and their labeled data … 19:07
Semi-supervised learning method using multi-task structural learning19:21