0.25
0.5
0.75
1.25
1.5
1.75
2
Supervised Clustering with Support Vector Machines
Published on Feb 25, 20078774 Views
Supervised clustering is the problem of training a clustering algorithm to produce desirable clusterings: given sets of items and complete clusterings over these sets, we learn how to cluster future s
Related categories
Chapter list
Supervised Clustering with SVMs00:01
Supervised Clustering Talk Outline00:21
Clustering Marbles00:38
Multiple Possible Criteria for Similarity00:52
How to Adjust01:02
Noun Phrase Coreference02:11
News Story Clustering02:48
Supervised Clustering Talk Outline03:16
How Do We Learn?03:31
Simple Clustering03:42
Clustering Objective04:13
Pairwise Features & Similarity05:05
Naïve Training Example05:54
Problem 1: Hard Coded Performance Measure06:40
Problem 2: Clustering Interactions07:27
Supervised Clustering Talk Outline08:15
SVMstruct Overview08:18
Ψ for Clustering09:20
Δ for Clustering10:05
Linear Constraint11:05
Quadratic Program Formulation11:56
Algorithm to Select Constraints12:27
Computing the Argmax13:22
Supervised Clustering Talk Outline15:12
NP Coreference15:21
News Story Clustering16:15
Building the pairwise feature vector ϕ16:45
SVMcluster vs. PCC16:58
Optimizing to Right Δ19:34
Inclusion of Δ in Finding Constraint20:57
Real Relaxation versus Greedy Clustering22:20
Conclusions23:30