In this practical you will use jupyter notebooks within Google Colab and learn how to use GPUs for deep learning.
This practical session explores how you can use neural networks. First you will examine how neural networks can be applied to tabular data, then you will explore convolutional neural networks for image classification.
The goal of this practical session is to show you that neural networks are much easier to code than to understand, and even simple networks can be very powerful.
What is in this Practical Session
- Neural Networks for Diagnosis of Diabetes
- Convolutional Neural Networks for Image Classification
It is suggested to read the notebooks in the above order. Note that exercises will appear within the notebooks this time, so that you don't need to code everything from scratch. You may find it beneficial to go through the notebooks without doing the exercises first, then come back to them.
Set up your notebook
We are not using binder for neural networks because it is too slow. Open up GoogleColab and import the notebooks in this repository to get started. You can do this by clicking File -> Open Notebook -> Github and search for KF5012-AI2021.
There is a lot of information about how to use Keras for different machine learning tasks. I particularly recommend Francois Chollet's repository (Francois Chollet is the founder of Keras).
Google Colab have a large number of tutorials on how to use Colab for different applications at https://research.google.com/seedbank/. You may look to find one similar to your project to play around with.