Uncertainty-aware Model-based Policy Optimization

Tung-Long Vuong, Kenneth Tran

Model-based reinforcement learning has the potential to be more sample efficient than model-free approaches. However, existing model-based methods are vulnerable to model bias, whi…