NeuraForge an easy neural network library

NeuralNet Class

The NeuralNet class is a basic implementation of a feedforward neural network. It has methods to set the weights of the network, forward propagate an input, and print the weights and biases.

Initialization

To initialize a NeuralNet object, the following parameters can be passed:

• input (int): Number of input neurons in the network.
• output_layer (int): Number of neurons in the output layer of the network.
• hidden (list): List of integers representing the number of neurons in each hidden layer of the network. If not provided, the network will have no hidden layers.
• add_biases (bool): Whether to add biases to the network. If True, biases will be added to each neuron in the network.
• activation (function): The activation function to be used in the network. If not provided, the network will not apply any activation function.

Methods

random(self, x, y=0) This method generates random values for the weights and biases of the network.

• x (int): Number of columns in the generated matrix.
• y (int, optional): Number of rows in the generated matrix. If not provided, a single row matrix is generated. Returns a list of Value objects.

setWeights(self) This method sets the weights and biases of the network based on the number of input and output neurons, as well as the number of neurons in any hidden layers.

dotproduct(self, layer, weights) This method calculates the dot product of a layer with a weight matrix.

• layer (list): The layer to be multiplied.
• weights (list): The weight matrix to be multiplied. Returns a list of dot products.

forward(self, x) This method performs a forward pass through the network with the given input.

• x (list): The input to be propagated through the network. Returns a list of Value objects representing the output of the network.

printWeights(self) This method prints the weights of the network.

printBiases(self) This method prints the biases of the network.

Attributes

• input (int): Number of input neurons in the network.
• output_layer (int): Number of neurons in the output layer of the network.
• hidden (list): List of integers representing the number of neurons in each hidden layer of the network.
• add_biases (bool): Whether to add biases to the network. If True, biases will be added to each neuron in the network.
• activation (function): The activation function to be used in the network. If not provided, the network will not apply any activation function.
• out (Value): A Value object representing the output of the network.