This folder contains Neural Network, CNN, RNN architectures made by Archit Naik and Melih Ekinci as course project in FAU. Architectures are made using NumPy.
Neural Network architecture consist of
a basic optimizer (SGD),
a base layer,
a fully connected layer,
ReLU and SoftMax Layers as activation functions,
and cross entropy loss function.
On top of the structures in NN, CNN architecture consist of
a set of initializers (Constant, UniformRandom, Xavier and He),
advanced optimizers (SGD with Momentum and Adam),
Flatten, convolution and pooling layers.
RNN in process..