Back to Course
Reinforcement Learning: Agents
Module 9 of 11
9. Curriculum Learning
1. Start Simple
You don't teach a child Calculus before Algebra. Don't ask an RL agent to solve the Hardest level immediately.
2. The Strategy
- Train on Easy environments (Reward is dense).
- Slowly increase difficulty (Reward becomes sparse).
- Result: Faster convergence and better generalisation.