Thesis ¶
Models ¶
For each of the 8 datasets you can find 10 trained baseline models and 10 trained optimized models here: zenodo.org/record/582892
Code ¶
To be fixed ¶
- STL-10 evaluation setup: I used the complete training set for training, but the datasets page gives 10-folds. Hence I have about 11% more labeled training data than intended.