This is a walkthrough of building a convolutional neural network (CNN) model to predict labels of the CIFAR-10 dataset. The contents are split in two Google Colab notebooks
- 0_Simple_Model.ipynb
- 1_Improved_Model.ipynb
The first model guides the reader through a simple setup of a CNN and adds regularization after evaluating results. The second model illustrates how the model's accuracy can be further increased by adding features such as data augmentation, a more complex network architecture, batch normalization, and dropout.
It is recommended to run both notebooks in Google Colab.