CM2003 Lab 3

In this lab, you will be exploring the following topics:

• Convolutional Neural Networks

• Regularization techniques: dropout, weight decay, data augmentation

• Transfer learning

• Hyperparameter tuning

You will be using the OrganSMNIST dataset.

Getting started

1. Start with the Convolutional Architectures notebook. This notebook will guide you through the process of training a convolutional neural network on the OrganSMNIST dataset. Convolutional Architectures

2. Continue with the Improving training and transfer learning notebook. This notebook will guide you through the process of improving the performance of your model using regularization techniques, weight decay, data augmentation, and transfer learning. Improving training and transfer learning

3. Finally, explore the different models and techniques we have provided you with. You can use the Model playground notebook to test out different models and techniques. Model playground

Submission

Answer the questions in the notebooks in a separate file and submit it on Github with your code. Please also write a short overview of the results and your observations.