DNN-classifire-using-TensorFlow_2.0

This is neural network with one input layer, two hidden layers and ten output layers. It uses TensorFlow2.0 and Keras as arhetecture of neural network.

How it works ?

At first il loads MINST Fashion dataset from Tensorflow and split it into train and test labels. Model uses Adam optimizer and saves trained model evry epoch in checkpoint directory that creates at the begining. Network has 6 layers:

1. First layer is input layer that reshapes input in 28x28 pixels
2. Second layer is first hidden layer with 200 neurons, it uses ELU activation function and Variance scaling initalizer
3. Third layer is Dropout function
4. Fourth layer is second hidden layer, it's preatty same as first hidden layer but has only 100 neurons
5. Fift layer is Batch normalization
6. Last layer is output layer that use Softmax activation function Model uses Adam optimizer and saves trained model evry epoch in checkpoint directory that creates at the begining.

How can I use it?

To use model you need installed Python and TensorFlow minimal version 2.0