End-To-End Learning AI

The goal of this project will be to implement an end-to-end learning artificial intelligence that is able to outperform human players in simulated environments, e.g. Atari games. This will be done with a reinforcement learning approach that is given only raw visual input from the simulated environment. The reinforcement approach can use neural networks as well as other, more classical learning algorithms. The different approaches should be implemented and evaluated against each other and of course the humans performance. Finally the whole system is supposed to run on a distributed cluster system, like the Hadoop framework.

For detailed information, please contact: M.Sc Matthias Wissel

Results

The results of this project were summarized in a presentation and explained in detail in the final documentation