In some cases, it is hard to design an appropriate reward function for a specific aim in the reinforcement learning environment.
At the seminar, we will discuss an approach of implicit reward design that uses expert-defined trajectories. We will present experimental results in Atari games and MuJoCo simulator.
Speaker: Mikhail Shavkunov.
Presentation language: Russian.
Date and Time: April 9th, 18:30-20:00.
Place: Times, room 204.
Videos from previous seminars are available at http://bit.ly/MLJBSeminars