Airhockey RL hackathon

Are you going?

22 people going

Share:
Location image of event venue

Details

IMPORTANT: We have postponed the hackathon to the weekend of 8-10 november

After the astrohack & music hackathon, it's time for our third hackathon :-)! This time, you will be able to make your own reinforcement learning agent that can play airhockey/pong!

Reinforcement learning agents that can beat human players in games have been in the news a lot of times. Wouldn't it be great if you could make one yourself? This hackathon is all about that :-)!
No need to be an airhockey champion or RL expert. We will start the hackathon by explaining the context & will also provide you with code for a minimal RL agent that is able to play the airhockey game. If you want to get some basics about reinforcement learning, certainly checkout these talks:
- https://www.youtube.com/watch?v=JgvyzIkgxF0
- https://www.youtube.com/watch?v=5P7I-xPq8u8
and relevant blogpost:
- http://karpathy.github.io/2016/05/31/rl/

You are welcome the weekend of November 9 at DigitYser. The hackathon ends on Sunday. The winner will be the group that can create the RL that is able to beat all the other agents.

Agenda:
- Friday 08/[masked]: opening hackathon and forming teams
- Friday 08/[masked]: explanation of problem and walkthrough to the sample code
- Friday 08/[masked]: official start hackathon
- Sunday 10/[masked]: end of hackathon
- Sunday 10/[masked]: ranking of the teams & live combat between the finalists to determine the winner
- Sunday 10/[masked]: end of award ceremony

FINAL Registration be required via a link on eventbrite, link will be shared soon! We will provide food, so we ask for a minimal contribution to cover our costs.