The key of this classifier, a stacked ensembling technique, was inspired from van der Laan's paper on super leaner.
The super learner is implemented based on the base estimator of sci-kit learn platform, so that this classifier can be compatible with sci-kit learn's modules, like metrics and grid-search.
More details can be found in the jupyter notebook and the assignment instructions.
Fashion MNIST was used as the dataset.
Below is the super learner's structure.
