en
0.25
0.5
0.75
1.25
1.5
1.75
2
Relevant Overlapping Subspace Clusters on Categorical Data
Published on Oct 07, 20142301 Views
Clustering categorical data poses some unique challenges: Due to missing order and spacing among the categories, selecting a suitable similarity measure is a difficult task. Many existing techniques r
Related categories
Chapter list
Relevant Overlapping Subspace Clusters on Categorical Data00:00
Motivation00:23
Unique challenge for categorical data00:28
Redundancy and parameters challenge - 100:51
Redundancy and parameters challenge - 201:31
Redundancy and parameters challenge - 301:56
Redundant Overlapping Clusters - 102:20
Redundant Overlapping Clusters - 202:35
Questions to be answered in this work03:01
Optimization Goal Compression03:36
Basic Idea03:40
Compress data modeled by clustering - 104:08
Compress data modeled by clustering - 204:30
Compress data modeled by clustering - 304:39
Compress data modeled by clustering - 405:09
Example of using compression to evaluate clustering05:33
Algorithm ROCAT05:54
Minimum Coding Problem is NP-hard05:59
Finding large pure subspace clusters first06:54
Searching phase - 107:12
Searching phase - 207:23
Searching phase - 307:27
Searching phase - 407:38
Searching phase - 507:40
Searching phase - 607:47
Searching phase - 707:50
Combining phase08:09
Reassinging phase08:21
Complexity08:44
Experiments09:07
Synthetic data sets09:14
Cluster Quality09:46
Subspace quality10:25
Robustness against outliers10:30
Results on real data10:55
Splice-junction Gene Sequences Data Set11:14
Results on Splice12:00
Summary12:34
Contributions12:39
Thank you!13:34