Java NeuralNetwork

I created with Java from scratch a deep feed forward neural network with gradient descent, and trained it on mnist and to do autoencoding, which worked well.

Targeted to extend the project for n dimensional convolutions (using tensor like data vectors), addition of cost functions, activation, error correction functions and LSTM/recurrent network, but abandoned since 2019.

I did also a bit of genetical algorithms for others tests during this time.