Posts
-
Summary - 95.28% on the test set
-
Batch normalization - 95.9%
-
Regularization - 94.9%
-
VGG, stride, and data augmentation - 94.8%
-
On network architecture and variance of results - 87.87%
-
Data processing revisited - 83.6%
-
Fixed bugs and increased the batch size
-
Analysis of my first attempt
-
First result: 71.5% accuracy!
-
Don't use test set for Cats&Dogs in Fuel!
-
Let's get it started!
subscribe via RSS