Reinforcement learning

In Reinforcement learning, learning is driven by interactions with an environment. Here an agent learns to behave in an environment by trial and error. Agent receive some rewards for taking actions that lead to desired outcomes and punishments for taking actions that lead to undesired outcomes.

 

Usecases where Reinforcement learning is used

1. Playing games
2. Self-driving cars
3. Algorithmic trading
4. Resource management
5. AlphaGo computer program designed using Reinforcement learning to play the game of Go.
6. Help robots to perform tasks in real world

 

Key terms in Reinforcement learning

1. Agent: Agent learns to behave in the environment, it can be a software program, robot. Agent makes decisions to maximize the rewards.
2. Environment: It is the system or place with which agent interacts with. Environment provides feedback to the agent based on its actions.
3. State: Represent the current state of the environment.
4. Action: It is the decision or some action made by agent. For example, moving up, raise arm etc.,
5. Reward: Reward is given to the agent when the correct action is taken

 

Previous                                                    Next                                                    Home