University of Twente Student Theses
Optimizing Crop Type Mapping for Fairness
Gorbunov, Ilya (2024) Optimizing Crop Type Mapping for Fairness.
|
PDF
1MB |
| Abstract: | This study investigated improving fairness in crop type mapping by applying fairness optimization methods to address both class and sensitive attribute imbalances using a transformer classifier. Parcel size was identified as the key sensitive attribute. The methods tested included Random Oversampling (RO), Weighted Cross Entropy (WCE), Focal Loss (FL), and two novel approaches that targeted class imbalance and the imbalance between small and large parcels: Random Oversampling with Resampling (RO-R), which increased representation of smaller parcels by redistributing random samples, and Double Objective Weighted Cross Entropy (DOWCE), which applied higher penalties to misclassification of smaller parcels. Hybrid methods, RO-DOWCE and RO-FL, were also evaluated. These methods were tested on diverse datasets subsampled from the BreizhCrops dataset, to assess their generalizability under varying conditions. Results showed RO-DOWCE was most effective at addressing class imbalance across the datasets, though not significantly different from RO-R, RO, and RO-FL. Cost-sensitive methods were generally less efficient compared to sample balancing and hybrid approaches. While all methods improved performance for both small and large parcels, reductions in the disparity between the two groups were marginal. |
| Item Type: | Essay (Master) |
| Faculty: | ITC: Faculty of Geo-information Science and Earth Observation |
| Subject: | 38 earth sciences |
| Programme: | Spatial Engineering MSc (60962) |
| Link to this item: | https://purl.utwente.nl/essays/104460 |
| 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