What is it about?
With Reinforcement Learning (RL), machines learn optimal behavior by interacting with environments, receiving rewards or penalties, and improving over time. It is a key driver of modern AI success in for example robotics and games.
History
RL has progressed to a core area of modern AI. Early milestones (80s) include Q-learning. Later advances with deep nets have enabled real-world applications, making RL a powerful tool for sequential decision-making in complex environments.
Example applications
- Robot motion control, e.g. locomotion
- Vehicle route optimization
- Board games and video games AI
- Coordination of autonomous drone fleets
- Renewable power system control
Why should you care?
There is little doubt that AI and RL are evolving fields set to revolutionize science and technology. The question you should ask yourself is: do you want to use a manual screwdriver while others have a far more efficient electric one?
