Training and testing the model
The following are the steps for training and testing the model:
- The first step is to read the training and testing datasets. Here are steps we must implement for reading the data:
- First, we load the training/testing images and label data from the files we downloaded for the Fashion MNIST data (https://github.com/zalandoresearch/fashion-mnist).
- Then, we reshape the image data to a shape of 28 x 28 x 1 for our model and normalize it by 255 to keep the input of the model between 0 and 1.
- We split the training data into train and validation datasets, each with 55,000 and 5000 images respectively.
- We convert our target array y for both training and testing datasets so that we have a one-hot representation of the 10 classes in the dataset that we are going to feed into the model.
Note
Make sure to choose around 10% of data for out validation. In this project, we choose 5000 random images (8% of the total images) for the validation data set.
The code for the preceding...
