You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
This project was created to test the ability of Siamese neural networks to correctly class data when class wasn't present in training set. Test Plans:
1. Build a siamese neural network and train on whole MNIST N training set. Measure performance of network for each class
2. Create N-1 training sets which have 1 class removed from dataset. Measure performance of network for all classes. Compare performance to the "N training set" and compare performance of the networks on the N-1 classes to the Nth class
3. Create N-2 training sets which have 2 classes removed. Measure performance of networks for all classes. Compare performance to the "N training set" and to the N-1 Training sets.