What is Deep Reinforcement Learning? How to implement it? Can the machine beat humans at games like poker?
In 2013, Deepmind blew everyone’s minds by being able to efficiently achieve superhuman ability in a range of atari games with only one algorithm. A few years later, AlphaGo beat the best players in the world at the game of go. the algorithms have never been taught about any of the rules of the games. All this was made possible by Deep Reinforcement Learning, which allowed the computer to play against itself billions of times to find the best strategy.
In this presentation, we will learn about the theory of Reinforcement Learning, which is one of the most exciting form of Machine Learning. We will then try to transpose and apply that to the game of poker. It will involve doing correct feature engineering to feed to the algorithm, training a model while taking the computational limitations into account, and benchmarking the results to what is a known ‘good’ strategy.
My name is Oscar Blazejewski. I’m a Data Engineer working at Adaltas. I am currently working for Renault, migrating older projects to their new HDP 2.6 platform.
I like all that is linked to technology, from Big Data to space travel, and love to learn new things.