bravotty/DataMining_AritificialNeuralNetworks

An Implementation of ANN

DataMining_AritificialNeuralNetworks

A python implementation of AritificialNeuralNetworks - ANN
Env : Python2.6

Usage :

PC python :

pip install numpy, pandas

Run .py

python Network.py

Defination :

AritificialNeuralNetworks Struct

class AritificialNeuralNetworks(object):
    def __init__(self, layers, learningRate, trainX, trainY, testX, testY, epoch):
        # input params
        self.layers   = layers 
        self.lr       = learningRate
        self.epoch    = epoch
        self.mean     = [np.mean(i) for i in trainX.T]
        self.stdVar   = [np.std(i)  for i in trainX.T]
        self.trainXPrediction = trainX
        self.trainYPrediction = trainY
        self.testXPrediction  = testX
        self.testYPrediction  = testY
        self.trainX   = self.Normalization(trainX)
        self.trainY   = self.oneHotDataProcessing(trainY)
        self.weights  = [np.random.uniform(-0.5, 0.5, [y, x]) for x, y in zip(layers[:-1], layers[1:])]
        self.biases   = [np.zeros([y, 1]) for y in layers[1:]]
        self.cntLayer = len(self.layers) - 1
        self.error    = None

Code Flie :

AritificialNeuralNetworks.py
  |--Initial params struct
  |--FitTransform Function
  |--ForwardUpdate Function
  |--BackForwardUpdate Function
  |--/* Activation function 
  		sigmoid - sigmoidPrime
  		tanh    - tanhPrime
  		ReLU    - ReLUPrime
  |--*/
  |--Cost Function
  |--Normalization Function
  |--OneHotDataProcessing Function
  |--Prediction Function
  |--Main

tools.py 
  |--CreateDataSet Function

License

The MIT License