University of Twente Student Theses
Active Learning Strategies for Long Horse Tails
De Jong, H.F. (2024) Active Learning Strategies for Long Horse Tails.
|
PDF
5MB |
| Abstract: | Recognizing symptoms of animal disease with machine learning can be improved by balancing class distributions of animal activity, which frequently are long-tailed. Uncertainty Sampling and Disagreement-based Sampling strategies of active learning, as well as Density Weighting and a novel Pragmatic Balance approach are evaluated on their resulting class distributions in this research. This is done by applying them to a dataset of horse accelerometer data. A combination of these approaches is shown to have a significant effect in achieving a balanced training set, by finding more instances of rare tail classes and reducing the amount of instances of common head classes in the training set. Additionally, general model performance increases noticeably with these methods. |
| Item Type: | Essay (Bachelor) |
| Faculty: | EEMCS: Electrical Engineering, Mathematics and Computer Science |
| Subject: | 54 computer science |
| Programme: | Computer Science BSc (56964) |
| Link to this item: | https://purl.utwente.nl/essays/100963 |
| Export this item as: | BibTeX EndNote HTML Citation Reference Manager |
Show download statistics for this publication
Show download statistics for this publicationRepository Staff Only: item control page