An Under the Hood Tune Up with AWS DeepRacer
Reinforcement learning (RL) has traditionally had a high barrier to entry within the industry. In this episode, we explore what happens under the hood while highlighting how AWS has democratized RL through the AWS DeepRacer service. Dr. Sahika Genc takes us through the interaction of the underlying core services of Amazon SageMaker and AWS RoboMaker as well as the algorithm and neural network architecture used by the agent while building the most optimized policy network through its feedback and experiences in simulation. We also touch on addressing one of the biggest challenges within the industry referred to as the reality gap and how the open sourced Amazon SageMaker sample notebook can be modified to create a good model.

View on YouTube