EDS Hackathon: Recurrent Neural Networks

Dear data enthusiasts,

Let's meet again for a hackathon and learn about Recurrnet Neural Networks

As we have mentioned during previous hackathons, we aim to build up the complexity of our hackathons over time. In that way we will be able to solve really interesting problems.

This hackathon is the third of a series of hackathon covering subjects in Neural Networks and Deep Learning.

In the previous two hackathons we have seen how we can build a Neural Network in Tensorflow, and how we can build a Convolutional Neural Network (CNN) and classify images.

This hack will focus on implementing a Recurrent Neural Network (RNN). A RNN differs from a CNN, because it can take data with variable size as input and therefore it is ideal to analyze time-series, signals or text documents.

We will build on the knowledge of the previous Hackathons, but I'll make this Hackathon as self-contained as possible. People who have missed Hackathon 8/9 should be able to complete this Hackathon without much difficulties.

Click here (https://github.com/RobRomijnders/EDS/tree/master/hack_11) for the GitHub repository.

Agenda:

18:00 Walk in
18:15 - presentation on RNN

• From CNN to RNN
• Different types of RNN
  18:45 Start hacking on assignments and your project
  20:45 Wrapup
  21:00-21:30 drinks

Note: the dataset will be CPU friendly:) We want to spend our time on learning, so we pick a cpu-friendly dataset.

Location:
The location (http://www.pluginpaviljoen.nl/fotos/) has a bar which we host ourselves. There will be coffee/tea and beer/soda. Please take some cash money to pay for your drinks.

https://secure.meetupstatic.com/photos/event/7/e/b/8/600_460652440.jpeg

PS: You can see the contents of the previous Hackathons here (https://github.com/RobRomijnders/EDS)