0.25
0.5
0.75
1.25
1.5
1.75
2
FastXML: A Fast, Accurate and Stable Tree-classifier for eXtreme Multi-label Learning
Published on Oct 08, 20143527 Views
The objective in extreme multi-label classification is to learn a classifier that can automatically tag a data point with the most relevant subset of labels from a large label set. Extreme multi-label
Related categories
Chapter list
FastXML - 100:00
Extreme Multi-Label Learning - 100:26
Extreme Multi-Label Learning - 200:54
Ranking and Recommendation - 101:17
Ranking and Recommendation - 201:46
Applications –Query Recommendation02:17
Applications –Video Recommendation02:52
MLRF and LPSR –Advantages03:05
MLRF and LPSR –Shortcomings03:51
FastXML - 204:15
Tree Based Extreme Classification05:00
FastXMLOverview05:27
Node Partitioning in Feature Space - 106:23
Node Partitioning in Feature Space - 206:35
Node Partitioning –Initialization06:43
Node Partitioning –Label Ranking - 107:05
Node Partitioning –Label Ranking - 207:10
Node Partitioning –Instance Assignment - 107:30
Node Partitioning –Instance Assignment - 207:46
Node Partitioning –Instance Assignment - 307:48
Node Partitioning –Label Ranking - 307:57
Node Partitioning –Instance Assignment - 408:00
Node Partitioning –Feature Space Partition08:08
Node Partitioning - 108:27
Node Partitioning - 208:34
Node Partitioning - 308:39
Node Partitioning - 408:44
Node Partitioning - 508:47
Data Set Statistics08:49
Results on Small Data Sets09:28
Large Data Sets -WikiLSHTC10:14
Large Data Sets –Ads-430K10:37
Large Data Sets –Ads-1M10:45
Training Time & Speedup vs. Cores10:53
Multiple Iterations -Ads-430K11:14
Tree Balance11:31
Variants of FastXML-Small Data Sets11:47
Variants of FastXML-Large Data Sets12:07
Random Tree Selection12:11
Conclusions12:25