Neural Network Steps

Original article was published on Deep Learning on Medium

Neural Network Steps

Single or multiple training examples.

  1. Define neural network structure
  2. Define parameter(w, b)or initialize
  3. For loop

1)define forward propagation

2) compute cost function
3)define backward propagation

4)update parameters using Gradient descent.

Do until minimize the cost function

After all

If y^ is less than 0.5 then output is zero


Y^ is greater than 0.5 then output is 1

In the neural network use single training example loss function =1/2(y-y^)

And cost function for multiple training example

L(y,y^)=1/m(y.log(y^[i])-(y-1).log (y^[i]-1)

In the hidden layer use tanh() activation function and output layer use sigmoid activation function .

In the backward propagation calculate the derivative function….

#machine learning#data science