Principled reward shaping for reinforcement learning via lyapunov stability theory.
Yunlong DongXiuchuan TangYe YuanPublished in: Neurocomputing (2020)
Keyphrases
- reward shaping
- reinforcement learning
- complex domains
- reinforcement learning algorithms
- numerical simulations
- markov decision problems
- state space
- chaotic systems
- model free
- lyapunov function
- function approximation
- multi agent
- control law
- adaptive control
- machine learning
- temporal difference
- optimal control
- markov decision processes
- sufficient conditions
- dynamic programming
- partially observable
- adaptive fuzzy
- agent learns
- control strategies