About this document
Nonlinear Preferences in Multi-Objective RL by 23020082 is a document available to read on EtoBox.
This document presents a study on multi-objective reinforcement learning (MORL) that focuses on maximizing expected scalarized return (ESR) using nonlinear preferences over trajectories. The authors derive an extended Bellman optimality principle for nonlinear optimization and propose an approximation algorithm for computing non-stationary policies, demonstrating its effectiveness through theoretical proofs and empirical results. This work is significant as it provides the first provable guarantees for appr
- Author
- 23020082
- Language
- EN