Deep Neural Network from scratch, scikit-learn, keras&Tensorflow part1

Basic

building neurons => feedforward => Loss(MSE) => backpropagation => Stochastic Gradient Descent => training => predict

  • ANN Model:

img

  • MSE:

mse

  • Backpropagation calculation:

back_1

back_2

  • Stocastic Gradient Descent:

SGD

Source code, deep neural network implementation from scratch, link here

PREVIOUSSoftmax function
NEXTPython Basic Data Structure