captain-pool / FC_FeedForward_SGD

Fully connected Feed Forward Neural Network Model using numpy. Stochastic Gradient Descent Algorithm

Fully Connected Feed Forward Neural Network

Overview

Lightweight Python 3 Library for Feed Forward Neural Network with Stochastic Gradient Descent (SGD) Optimizer and Euclidean Error Function.

Dependencies:

• Numpy 1.14.xpip install numpy

Usage:

• Create Network

from FC_SGD import Network
    
import numpy as np
network=Network([3,5,2]) # Fully Connected Neural Network created with 3 layers, having 3 neurons, 5 neurons, and 2 neurons

• Train Network

network.SGD(np.array([([<input layer data 1>],[output layer data 1]), ([<input layer data 2>],[output layer data 2])]))

• Predict result

output=network.feedforward(np.array([<input layer data>]))
print(output[<last layer neuron number 0-base indexed>]) #0th Neuron is at the top of the layer