Scalable Oversight
Machine Learning
An on-policy learning algorithm that updates Q-values based on state-action-reward-state-action transitions.
Detailed Explanation
The SARSA Algorithm is an on-policy reinforcement learning method used to estimate optimal action-value functions. It updates Q-values based on observed transitions, specifically considering the current state-action pair, the reward received, and the next state-action pair. This approach enables the agent to learn policies that maximize cumulative rewards by exploring and exploiting actions simultaneously.
Use Cases
- •Use case: Teaching a robot to navigate a maze by iteratively updating its path choices based on immediate rewards and subsequent actions.
Related Terms
Other terms in the Machine Learning category