mrucker / kpirl-kla Goto Github PK

This repository contains two new algorithms: KPIRL and KLA. KPIRL is a non-linear extension to Abbeel and Ng's Projection IRL algorithm (detailed in "Apprenticeship Learning via Inverse Reinforcement Learning"). KLA is an approximate RL algorithm designed to be used with KPIRL in large state-action spaces without any reward shaping. The algorithms have been published in "Human Apprenticeship Learning via Kernel-based Inverse Reinforcement Learning."

Home Page: https://arxiv.org/abs/2002.10904

License: MIT License