Skip to content

Zahra-Bakhtiari/Building-Deep-Neural-Network-Step-by-Step-Instruction

Repository files navigation

The first notebook Planar_data_classification_with_one_hidden_layer in this repository covers a simple implementation of a neural network, which will only have one hidden layer. This notebook covers below steps:

  1. Implement a 2-class neural network with a single hidden layer

  2. Use units with a non-linear activation function, such as tanh

  3. Compute the cross entropy loss

  4. Implement forward and backward propagation

The second notebook Building your Deep Neural Network Step by Step in this repository takes it one step further and explains steps to build a deep neural network with as many layers as we want! The model highlights include:

  1. Using non-linear units like ReLU to improve your mode
  2. Building a deeper neural network (with more than 1 hidden layer)
  3. Implementing an easy-to-use neural network class which can easily be integrated into an OOP code base for building complex services.

Releases

No releases published

Packages

No packages published