Applying Deep Convolutional Layers on the popular MNIST dataset for recognising Hand-written digits.
Will be experimenting with various layer architectures to come up with the best accuracy.
Best Model as of now --
2 Conv2D layers + MaxPool + 2 Conv2D layers + Max pool + 2 Dense Layers (with a dropout)
Accuracy -- 98.94%
MNIST Dataset Source -- Kaggle