Offline Deep Reinforcement Learning

[Expired]

Introduction

Reinforcement Learning (RL) is a class of machine learning algorithms in which an agent interacts by trial-and-error in an environment [1]. Together with Deep Learning, RL has obtained excellent results in a great number of simulated environments, like video-games [2] or board-games, or even robot manipulation. 

The majority of Deep Reinforcement Learning (DRL) algorithms developed so far use an Online learning setting, in which the agent directly interacts with the environment in order to collect data and gradually improve its behavior. This approach limits the adoption of RL algorithms in industrial applications, because deploying a not trained agent in a real environment can take a prohibitive amount  of time and can also be dangerous.

For these reasons, the research community has proposed an offline variant called Offline RL [3,4] (also known as Batch RL), in which the agent learns from a fixed and provided dataset [5].

This thesis aims to deepen the Offline RL approaches [6,7], in order to investigate the possibilities and limits of solutions proposed comparing them in a benchmark environment.

Planned Activities

The planned activities for this thesis are:

  1. Acquire strong theoretical basis on Deep Reinforcement Learning;
  2. Investigate the different Offline DRL approaches;
  3. Compare Offline DRL algorithms in a benchmark environment.

Who we’re looking for

Students that are about to get their Master Degree in: computer science, computer engineering, mechatronic engineering, mathematical engineering, mathematics, physics, informatics.

Required skills are:

  • Proficiency in at least one programming language (Python, C++), Python is preferred;
  • Basic knowledge of machine learning and Deep Learning algorithms;
  • Good mathematical, statistical and analytical skills

Duration of this Projects: 6-8 months

Check these links before moving on

Contact Us

Directly by email to: [email protected]

By LinkedIn: linkedin.com/in/cannavò-sonia-66a95467