Neural network basic

Neural network basic

• 1.
  What is aNeural Network? The term ‘Neural’ has origin from the human (animal) nervous system’s basic functional unit ‘neuron’ or nerve cells present in the brain and other parts of the human (animal) body. A neural network is a group of algorithms that certify the underlying relationship in a set of data similar to the human brain. The neural network helps to change the input so that the network gives the best result withoutredesigning the output procedure. Components of Artificial Neural Network: 1-InputValues 2-OutputValue (Desired/Expected) 3-InitialWeights 4-InitialBias (Default =1) (Feed Forward) 5-Pre-Activation Function(∑ 𝑥𝑖 − 𝑛 𝑖=1 wi+b) 6-Activation Function (sigmoid) 7-CurrentOutput 8-updateweights (Backpropagation)
• 2.
  Process: - Using gradientdescent Algorithm wherechain rule applied for Backpropagation,firstTakeinput Feature, they’re going to be an output values, Inputis multiplied with given weights(initial)+bias(default=1) and feed forward to pre activation function to calculate the sum of all the feeds, then sigmoid function and then calculate the desired output, the convergenceof weights needs Backpropagation the update the weights of each input (hidden layers) Explained in code: a) Input Layer NumPy Array b) DesiredOutputs c) Weights/Bias Initialization InputLayer: Use 4 valuesin matrix format DesiredOutput:Eachinput matrix has some desired output.It cannotinjectinto Neural Network ininputs only use whenerroror loss calculation Initial Weights associatedwith inputs Biasdefault value =1
• 3.
  d) ActivationFunction HIDDEN LAYERSCALCULATIONS e) Chain Rule in gradient descent, Update Of weights &Calculationof Pre- ActivationFunction Steep Activation Function Derivative calculationof Activation Function Epochs isa terminmachine learningindicatesthe numberof passesof entire trainingdataset----Noof Iterations Calculation of all weight’s pre- activation(∑ 𝑥𝑖 − 𝑛 𝑖=1 wi+b) function Calculationof Error and Losstotal error. Current output Update of Weights Backpropagation Update of Bias Backpropagation
• 4.
  f) Prediction Is onthe Basis of comparison between the calculated output and desired output The desiredoutputof input matrix [0,0] is0 The outputis appr to 0. The outputis appro The desiredoutputof input matrix [0,1] is1 The outputis appr to 1. The outputis appro The desiredoutputof input matrix [1,1] is1 The outputis appr to 1. The outputis appro The desiredoutputof input matrix [1,0] is0 The outputis appr to 0. The outputis appro