APReL is a unified Python3 library for active preference-based reward learning methods. It offers a modular framework for experimenting with and implementing preference-based reward learning techniques; which include active querying, multimodal learning, and batch generation methods.
Not sure what preference-based learning means? Read on or check our talk at AI-HRI 2021. You can also take a look at our video which contains a simple example.
overview installation example video Paper <https://arxiv.org/abs/2108.07259> Talk <https://youtu.be/HExrlibCxdI>
source/aprel
acknowledgement references
genindexmodindex