Name
Abstract | ||
|---|---|---|
In practical classification, there is often a mix of learnable and unlearnable classes and only a classifier above a minimum performance threshold can be deployed. This problem is exacerbated if the training set is created by active learning. The bias of actively learned training sets makes it hard to determine whether a class has been learned. We give evidence that there is no general and efficient method for reducing the bias and correctly identifying classes that have been learned. However, we characterize a number of scenarios where active learning can succeed despite these difficulties. |
Year | DOI | Venue |
|---|---|---|
2006 | 10.1145/1183614.1183709 | CIKM |
Keywords | Field | DocType |
minimum performance threshold,training set,practical text classification,active learning,practical classification,efficient method,unlearnable class | Training set,Data mining,One-class classification,Active learning,Computer science,Artificial intelligence,Thresholding,Classifier (linguistics),Learnability,Machine learning | Conference |
ISBN | Citations | PageRank |
1-59593-433-2 | 18 | 1.21 |
References | Authors | |
22 | 3 | |
Name | Order | Citations | PageRank |
|---|---|---|---|
| Hinrich Schütze | 1 | 2113 | 362.21 |
| Emre Velipasaoglu | 2 | 133 | 6.61 |
| Jan O. Pedersen | 3 | 6301 | 1177.07 |