Build your own neural network!

If you know what machine learning is then chances are you've heard of neural networks. You've certainly used them before - in recent years they have become a vital part of services offered by Google, Facebook and Apple. Perhaps you've even consciously used one, using pre-existing packages such as Tensorflow. But have you ever wondered what's "under the bonnet" of these mysterious neural nets? Or how you'd go about making one yourself?

In just a couple of hours and starting from absolutely nothing, we'll each build a fully-connected neural network capable of recognising handwritten digits. In the process we'll discover exactly what's going on inside the network at every stage, pick up some tips for working with R, and enter a Kaggle competition!

This is very much intended as an interactive session ran by our member Owen Jones. If you can, then please bring a laptop with R and RStudio installed (no experience of R needed!) but otherwise do come along anyway and we'll be able to partner up. It will also save a bit of time if you download the data we'll be using in advance - see the link below. Hope to see you there!

Download R: https://cloud.r-project.org/
Download RStudio: https://www.rstudio.com/products/rstudio/download/
Download data (train.csv and test.csv): https://www.kaggle.com/c/digit-recognizer/data