An article by Parnian Barekatain in which she describes some basic concepts in Reinforcement Learning. She also provides you with the link to Udacity’s free course on Deep Learning with Pytorch.
The blog consists of four parts:
- Deep Neural Network for Single-Agent: Reinforcement Review, DQN and Replay Memory
- Overview of multi-agent Reinforcement Learning
- Deep Neural Network for multi-agent: Independent Q Learning (IQL) and challenges combing with multi-agent Reinforcement Learning
- FingerPrinting
In 2015, DeepMind was able to successfully combine Deep Neural Network with Reinforcement Learning for a single-agent. Combining Deep Neural Network with Reinforcement Learning enabled AI for the first-time to surpass the performance of professional human players across many game scenarios.
This worked well fora single agent. However, when we have many agents, we cannot easily combine Deep Neural Networks with Reinforcement Learning, mainly because each agent constantly changes the dynamic of the environment and makes it really hard for other agents to learn what to do.
Read the rest of the article and watch explanation videos to understand this paper. Plenty of links to other resources to get you on track! Excellent read for any aspiring data scientist.
[Read More]