Kernel Image Classification Challenge

By Emile Mathieu & Thomas Pesneau

Code for the Challenge: https://inclass.kaggle.com/c/kernel-methods-for-machine-learning-data-challenge

Numpy, pandas and scipy are required

Files 'Xtr.csv', 'Xte.csv' and 'Ytr.csv' must be in the same folder as 'start_cnn.py' or 'start_kmeans.py'. Then run python start_kmeans.py (yields ~58% in ~5-10mn)

python start_cnn.py (yields ~52% in ~5-10mn)

For best performance (yields ~62% but take ~30mn), run python start_kmeans.py best