Project by Dawn McKnight and Zeyi Wang in Winter 2020 for the CMPUT 659: Explainable AI course at the University of Alberta.
Artifical intelligences can be tailored for a variety of means, including performance, or how strong agents are; and explainability, or how human-comprehensible the resulting agents are. In this project, we explore the usage of an evolutionary algorithm equipped with a domain-specific language for the purpose of finding explainable, relatively strong artificial intelligence for the popular video game series Pokémon.
The report for our project can be found at https://github.com/dem1995/pokemon-evolutionary-agents/blob/master/Pokemon_Evolutionary_Agents_Report.pdf
A Jupyter notebook is provided for demoing the project. After cloning our repository, please refer to sandbox.ipynb.