0.25
0.5
0.75
1.25
1.5
1.75
2
Beyond Hartigan Consistency: Merge Distortion Metric for Hierarchical Clustering
Published on Aug 20, 20151679 Views
Hierarchical clustering is a popular method for analyzing data, which associates a tree to a dataset. Hartigan consistency has been used extensively as a framework to analyze such clustering algorithm
Related categories
Chapter list
Beyond Hartigan Consistency00:00
The goal of clustering00:07
In this talk ...00:30
What structure do we wish to recover?03:16
High-density clusters - 103:41
High-density clusters - 204:15
High-density clusters - 304:20
High-density clusters - 404:25
A hierarchy of clusters 04:38
The density cluster tree04:51
What structure do we wish to recover?05:21
Recovering the density cluster tree from data - 105:33
Recovering the density cluster tree from data - 206:00
Recovering the density cluster tree from data - 306:12
Recovering the density cluster tree from data - 406:27
Recovering the density cluster tree from data - 506:48
Recovering the density cluster tree from data - 607:06
Recovering the density cluster tree from data - 707:34
Recovering the density cluster tree from data - 807:44
Recovering the density cluster tree from data - 908:00
Recovering the density cluster tree from data - 1008:03
What properties ensure that an algorithm captures the density cluster tree?08:07
Hartigan Consistency -108:45
Hartigan Consistency - 208:59
Hartigan Consistency - 309:21
Hartigan Consistency - 409:24
Hartigan Consistency - 509:28
Hartigan Consistency - 609:32
Hartigan Consistency - 709:43
Hartigan Consistency - 809:48
Hartigan Consistency - 909:50
Hartigan Consistency - 1009:51
Hartigan Consistency - 1109:52
Hartigan Consistency - 1210:01
Hartigan Consistency - 1310:04
Hartigan Consistency - 1410:10
Hartigan Consistency - 1510:15
What properties ensure that an algorithm captures the density cluster tree? - 110:21
Hartigan Consistency is insufficient - 111:35
Hartigan Consistency is insufficient - 211:59
Hartigan Consistency is insufficient - 312:02
Hartigan Consistency is insufficient - 412:03
Hartigan Consistency is insufficient - 512:13
Hartigan Consistency is insufficient - 612:14
Hartigan Consistency is insufficient - 712:31
Hartigan Consistency is insufficient - 812:34
Hartigan Consistency is insufficient - 912:41
Hartigan Consistency is insufficient - 1012:47
Hartigan Consistency is insufficient - 1112:57
Beyond Hartigan consistency13:07
Minimality - 113:46
Minimality - 214:06
Minimality - 314:37
Minimality - 414:39
Minimality - 514:41
Separation - 114:54
Separation - 215:15
Separation - 315:17
Separation - 415:19
Separation - 515:21
What properties ensure that an algorithm captures the density cluster tree? - 215:44
Ideal and empirical merge height - 116:13
Ideal and empirical merge height - 216:30
Ideal and empirical merge height - 316:36
Ideal and empirical merge height - 416:37
Ideal and empirical merge height - 516:38
Ideal and empirical merge height - 616:46
Ideal and empirical merge height - 716:47
Ideal and empirical merge height - 816:49
Ideal and empirical merge height - 916:52
Ideal and empirical merge height - 1016:58
Theorem17:22
Minimality, separation and the merge distortion metric17:33
Convergence of robust single linkage17:55
Future work18:26
Summary - 118:33
Summary - 218:35
Thank you18:55