I just won the Dogs vs. Cats Kaggle competition, using the deep learning library I wrote during my PhD: OverFeat http://cilvr.nyu.edu/doku.php?id=code:start

My system was pre-trained on ImageNet (ILSVRC12 classification dataset) and subsequently refined on the cats and dogs data. I limited my number of submissions to 5 to avoid test set tuning and obtained 1st place against 215 other teams with 98.9% accuracy.
Shared publiclyView activity