The handwritten-digit-recognition-final.ipynb file consists of the python code comprising of the different experiments with learning rates, activation functions, number of neurons in the hidden layer and optimisers.
The plots can be found in /plots
The output files can be found in /outputs
The best results have been obtained for the SGD optimiser with learning rate of 0.001 and 128 neurons with 'Relu' as the activation