Settling the Reward Hypothesis

12/20/2022
by   Michael Bowling, et al.
4

The reward hypothesis posits that, "all of what we mean by goals and purposes can be well thought of as maximization of the expected value of the cumulative sum of a received scalar signal (reward)." We aim to fully settle this hypothesis. This will not conclude with a simple affirmation or refutation, but rather specify completely the implicit requirements on goals and purposes under which the hypothesis holds.

READ FULL TEXT

page 1

page 2

page 3

page 4

research
05/17/2021

Learning to Win, Lose and Cooperate through Reward Signal Evolution

Solving a reinforcement learning problem typically involves correctly pr...
research
06/12/2020

The United Nations Sustainable Development Goals in Systems Engineering: Eliciting sustainability requirements

This paper discusses a PhD research project testing the hypothesis that ...
research
08/11/2020

Identifying Implicit Vulnerabilities through Personas as Goal Models

When used in requirements processes and tools, personas have the potenti...
research
01/09/2023

On The Fragility of Learned Reward Functions

Reward functions are notoriously difficult to specify, especially for ta...
research
10/22/2019

Teach Biped Robots to Walk via Gait Principles and Reinforcement Learning with Adversarial Critics

Controlling a biped robot to walk stably is a challenging task consideri...
research
04/20/2022

Joint Learning of Reward Machines and Policies in Environments with Partially Known Semantics

We study the problem of reinforcement learning for a task encoded by a r...
research
07/07/2020

Training design fostering the emergence of new meanings toward unprecedented and critical events

Our research is part of a technological research program in adult educat...

Please sign up or login with your details

Forgot password? Click here to reset