Budde, Lina E. ; Bulatov, Dimitri ; Strauss, Eva ; Qiu, Kevin ; Iwaszczuk, Dorota
Hrsg.: Köthe, Ullrich ; Rother, Carsten ; German Association for Pattern Recognition (DAGM) (2024)
Characterization of Out-of-distribution Samples from Uncertainty Maps Using Supervised Machine Learning.
DAGM German Conference on Pattern Recognition. Heidelberg, Germany (19.09. - 22.9.2023)
doi: 10.1007/978-3-031-54605-1_17
Konferenzveröffentlichung, Bibliographie
Kurzbeschreibung (Abstract)
The quality of land use maps often refers to the data quality, but distributional uncertainty between training and test data must also be considered. In order to address this uncertainty, we follow the strategy to detect out-of-distribution samples using uncertainty maps. Then, we use supervised machine learning to identify those samples. For the investigations, we use an uncertainty metric adapted from depth maps fusion and Monte-Carlo dropout based predicted probabilities. The results show a correlation between out-of-distribution samples, misclassifications and uncertainty. Thus, on the one hand, out-of-distribution samples are identifiable through uncertainty, on the other hand it is difficult to distinguish between misclassification, anomalies and out-of-distribution.
| Typ des Eintrags: | Konferenzveröffentlichung |
|---|---|
| Erschienen: | 2024 |
| Herausgeber: | Köthe, Ullrich ; Rother, Carsten |
| Autor(en): | Budde, Lina E. ; Bulatov, Dimitri ; Strauss, Eva ; Qiu, Kevin ; Iwaszczuk, Dorota |
| Art des Eintrags: | Bibliographie |
| Titel: | Characterization of Out-of-distribution Samples from Uncertainty Maps Using Supervised Machine Learning |
| Sprache: | Englisch |
| Publikationsjahr: | 8 März 2024 |
| Ort: | Cham |
| Verlag: | Springer Nature Switzerland |
| Buchtitel: | Pattern Recognition |
| Reihe: | Lecture Notes in Computer Science |
| Band einer Reihe: | 14264 |
| Veranstaltungstitel: | DAGM German Conference on Pattern Recognition |
| Veranstaltungsort: | Heidelberg, Germany |
| Veranstaltungsdatum: | 19.09. - 22.9.2023 |
| DOI: | 10.1007/978-3-031-54605-1_17 |
| URL / URN: | https://doi.org/10.1007/978-3-031-54605-1_17 |
| Kurzbeschreibung (Abstract): | The quality of land use maps often refers to the data quality, but distributional uncertainty between training and test data must also be considered. In order to address this uncertainty, we follow the strategy to detect out-of-distribution samples using uncertainty maps. Then, we use supervised machine learning to identify those samples. For the investigations, we use an uncertainty metric adapted from depth maps fusion and Monte-Carlo dropout based predicted probabilities. The results show a correlation between out-of-distribution samples, misclassifications and uncertainty. Thus, on the one hand, out-of-distribution samples are identifiable through uncertainty, on the other hand it is difficult to distinguish between misclassification, anomalies and out-of-distribution. |
| Freie Schlagworte: | Potsdam dataset, error detection, semantic segmentation |
| Fachbereich(e)/-gebiet(e): | 13 Fachbereich Bau- und Umweltingenieurwissenschaften 13 Fachbereich Bau- und Umweltingenieurwissenschaften > Institut für Geodäsie 13 Fachbereich Bau- und Umweltingenieurwissenschaften > Institut für Geodäsie > Fernerkundung und Bildanalyse |
| Hinterlegungsdatum: | 12 Mär 2024 07:13 |
| Letzte Änderung: | 12 Mär 2024 07:13 |
| PPN: | |
| Export: | |
| Suche nach Titel in: | TUfind oder in Google |
 |
Frage zum Eintrag |
Optionen (nur für Redakteure)
 |
Redaktionelle Details anzeigen |
Drucken
Impressum