Tracking the race between deep reinforcement learning and imitation learning

DOI10.1007/978-3-030-59854-9_2zbMATH Open1483.68339arXiv2008.00766OpenAlexW3096351713MaRDI QIDQ2056979FDOQ2056979

Authors: Timo P. Gros, Daniel Höller, J. Hoffmann, Verena Wolf 

Publication date: 8 December 2021

Abstract: Learning-based approaches for solving large sequential decision making problems have become popular in recent years. The resulting agents perform differently and their characteristics depend on those of the underlying learning approach. Here, we consider a benchmark planning problem from the reinforcement learning domain, the Racetrack, to investigate the properties of agents derived from different deep (reinforcement) learning approaches. We compare the performance of deep supervised learning, in particular imitation learning, to reinforcement learning for the Racetrack model. We find that imitation learning yields agents that follow more risky paths. In contrast, the decisions of deep reinforcement learning are more foresighted, i.e., avoid states in which fatal decisions are more likely. Our evaluations show that for this sequential decision making problem, deep reinforcement learning performs best in many aspects even though for imitation learning optimal decisions are considered.

Full work available at URL: https://arxiv.org/abs/2008.00766

Recommendations

• scientific article; zbMATH DE number 2038829
• scientific article; zbMATH DE number 6433487
• scientific article; zbMATH DE number 1738682
• Imitation learning as \(f\)-divergence minimization
• From Reinforcement Learning to Deep Reinforcement Learning: An Overview

zbMATH Keywords

imitation learningdeep reinforcement learning

Mathematics Subject Classification ID

Artificial neural networks and deep learning (68T07) Problem solving in the context of artificial intelligence (heuristics, search strategies, etc.) (68T20)

Cited In (5)

• Title not available (Why is that?)
• Title not available (Why is that?)
• DSMC evaluation stages: fostering robust and safe behavior in deep reinforcement learning -- extended version
• Imitation learning of car driving skills with decision trees and random forests
• Title not available (Why is that?)

This page was built for publication: Tracking the race between deep reinforcement learning and imitation learning

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q2056979)