[Data challenge]

Presentation

This is the code for my various submissions at the data challenge [TO BE REVEALED]

Preprocessing the entire training set before cross-validation is slightly cheating - maybe implement a new wrapper class and use pipelines?
Subsample pre-computed KS features to enable usage

Name		Name	Last commit message	Last commit date
Latest commit History 32 Commits
_backup		_backup
deeplearning		deeplearning
.gitignore		.gitignore
README.md		README.md
bag_of_words.py		bag_of_words.py
benchmark.py		benchmark.py
classification.py		classification.py
cnn_1D.py		cnn_1D.py
distances.py		distances.py
exploration.py		exploration.py
isi_transformation.py		isi_transformation.py
neighbors_isi.py		neighbors_isi.py
preprocessing.py		preprocessing.py
results_exploration.py		results_exploration.py