Tutorial on Explainable Deep Reinforcement Learning: One framework, three paradigms and many challenges

Interpretability, explainability and transparency are key issues to introducing Artificial Intelligence closed-box methods in many critical domains: This is important due to ethical concerns and trust issues strongly connected to reliability, robustness, auditability and fairness, and has important consequences towards keeping the human in the loop in high levels of automation, especially in critical cases for decision making. Reinforcement learning methods, and especially their deep versions, are closed-box methods that support agents to act autonomously in the real world. This tutorial will provide a formal specification of the deep reinforcement learning explainability problems, and will present the necessary components of a general explainable reinforcement learning framework. Based on this framework will provide distinct explainability paradigms towards solving explainability problems, with examples from state-of-the-art methods and real-world cases. The tutorial will conclude identifying open questions and important challenges. The tutorial is based on the survey paper on "Explainable Deep Reinforcement Learning" State of the Art and Challenges" [1].