About this document
Tractable Solutions for Regularized IRL by hitzcy2016 is a document available to read on EtoBox.
The document discusses Regularized Inverse Reinforcement Learning (IRL), which aims to derive reward functions that enable learners to imitate expert behavior while avoiding degenerate solutions. It presents tractable methods for regularized IRL applicable to both discrete and continuous control problems, including a novel approach called Regularized Adversarial Inverse Reinforcement Learning (RAIRL). The authors empirically validate their methods across various tasks, demonstrating their effectiveness in p
- Author
- hitzcy2016
- Language
- EN