Intro to Deep Learning

Intro to Deep Learning

Deep Learning is a bit of a buzzword in the industry at the moment and is the state-of-the-art in AI, but what exactly is it? In this workshop we'll explain exactly that and by the end you'll be able to build your very own neural network from scratch - to predict housing prices and even detect whether cancer cells are benign/malignant!


Anyone who is familiar with programming in Python will be able to follow along.

Note: Google Colaboratory requires you to log into a Google account (so it can save the notebooks in Google Drive).

Set up instructions

No setup is required since we're hosting the notebook online - either clone the repo or just click the link in the .ipynb file to open up Google Colaboratory online:

Make sure you're logged into a Google account to use Colaboratory!

Alternatively, you could run the notebook locally, however you will have to install Jupyter Notebooks and the relevant libraries imported in the notebook (e.g. keras).

Contributors: Mukul Rathi
View code examples on GitHub
License: MIT

Getting started with the code:

Instructions for implementing the neural network are all in the Google Colaboratory notebooks in the repo.

To load the notebook, head to Google Colaboratory and click the "Upload" tab, (if you have cloned the repo) and select the Intro to Deep Learning.ipynb / Intro to Deep Learning2.ipynb file on your computer (for workshop 1 / 2 respectively). You can also load directly from GitHub if you click the GitHub tab and enter in the search field for workshop 1 and for workshop 2.

When you initially try to run a cell, there will be a "Warning, this notebook wasn't authored by Google" - uncheck the "reset all runtimes" box and click "run anyway".

Finally, to run the code you can just click the "play" button on each cell to execute the code in that cell.

If you get stuck, ask a demonstrator (there is also a fully-implemented notebook in the repo for reference).

Diving deeper: (no pun intended)

To read up more about the neural net we're coding up, there are a couple of blog posts on neural nets and backpropagation.

The blog also covers the content in the slides in more depth (and more maths, if you're interested in that).

Further reading