Name
Abstract | ||
|---|---|---|
•Open set domain adaptation (OSDA) as a positive-unlabeled (PU) learning problem.•Novel reconstruction-based risk estimator for PU learning, resilient to domain shift.•Novel OSDA algorithm that uses PU learning and domain adversarial training.•Evaluation metric for OSDA that balances the performance on known and unknown classes. |
Year | DOI | Venue |
|---|---|---|
2020 | 10.1016/j.patrec.2020.06.003 | Pattern Recognition Letters |
Keywords | DocType | Volume |
Computer vision,Deep learning,Image classification,Domain adaptation,Open set recognition,Positive-Unlabelled learning | Journal | 136 |
ISSN | Citations | PageRank |
0167-8655 | 0 | 0.34 |
References | Authors | |
0 | 3 | |
Name | Order | Citations | PageRank |
|---|---|---|---|
| Mohammad Reza Loghmani | 1 | 1 | 2.38 |
| Markus Vincze | 2 | 1343 | 136.87 |
| Tatiana Tommasi | 3 | 502 | 29.31 |