Interpretable Reinforcement Learning using Dynamic Decision Trees
Year: 2024
Type: dissertation
Abstract: The purpose of this work is to delve into the field of Interpretable Reinforcement Learning using dynamic decision trees. Reinforcement Learning is a learning method for an intelligent system or agent that learns to act in an environment through trial and error. It is one of the three categories of Machine Learning and has a wide range of applications in areas such as healthcare, robotics, economics, etc. Reinforcement Learning algorithms often lack interpretability. The ability to interpret these algorithms is important for various reasons, such as error detection, safety, and system trust. In this work, an analysis of the Conservative Q Improvement (CQI) algorithm is conducted, which utilizes dynamic decision trees to generate easily interpretable policies.
Author: Karabelos, George
Language: English
Subfield: Artificial Intelligence
Field: Computer Science
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W6930580478
Cites:
Cited by:
Related to: