en
0.25
0.5
0.75
1.25
1.5
1.75
2
Active learning with evolving streaming data
Published on 2011-11-293582 Views
Since labeling data is often both laborious and costly, the labeled data available in many applications is rather limited. Active learning is a learning approach which actively selects unlabeled data
Related categories
Presentation
Active learning with evolving streaming data00:00
setting00:50
Data stream mining00:52
Examples of data streams01:58
Mining data streams02:31
Predictive models for data streams - 103:40
Predictive models for data streams - 203:56
Predictive models for data streams - 304:05
Predictive models for data streams - 404:17
Active learning for data streams - 104:59
Active learning for data streams - 205:29
Active learning for data streams - 306:06
Problem setting summary06:38
Contributions07:11
active learning strategies for data streams07:45
Do we need this label? - 107:58
Do we need this label? - 208:30
Random strategy (naive) - 108:32
Random strategy (naive) - 209:25
Do we need this label? - 309:31
Fixed uncertainty10:16
Online active learning in the data stream setting?11:06
What is needed? - 111:33
Budget - 111:51
Budget - 212:15
Budget - 312:39
What is needed? - 212:43
Changes in the regions where classifier is very certain should not be missed13:26
Do we need this label? - 413:46
What is needed? - 313:53
Adaptive uncertainty strategy14:07
Randomized uncertainty14:52
empirical results15:38
MOA15:46
Experimental evaluation16:41
REUTERS data - 117:47
REUTERS data - 218:35
REUTERS data - 319:05
Conclusion - 119:35
Conclusion - 219:36
Thanks!20:13