Layers

class gwu_nn.layers.Layer(activation=None)

The Layer layer is an abstract object used to define the template for other layer types to inherit

class gwu_nn.layers.Dense(output_size, add_bias=False, activation=None, input_size=None)

The Dense layer class creates a layer that is fully connected with the previous layer. This means that the number of weights will be MxN where M is number of nodes in the previous layer and N = number of nodes in the current layer.

init_weights(input_size)

Initialize the weights for the layer based on input and output size

Args:

input_size (numpy array): dimensions for the input array