Generalisation

Reinforcement learning is often framed around episodic, discounted, or average scalar rewards. While useful, these views miss a core …

While reinforcement learning has achieved remarkable successes in several domains, its real-world application is limited due to many …