AI Paper Atlas

Research map / Multi-agent systems and games / Games and equilibria

On the Expressivity of Markov Reward

David Abel, Will Dabney, Anna Harutyunyan, Mark K Ho, Michael Littman, Doina Precup, Satinder Singh

NeurIPS 2021 · Multi-agent systems and games › Games and equilibria · approach: Learning with strategic agents · 0 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
Specify tasks as SOAPs, policy orderings, or trajectory orderings and require Markov reward to realize themFinite CMPs · SOAPs, POs, and TOstask expressibilityMixed
Use range-SOAP instead of equal-SOAP realizationMarkov reward design · SOAP realizationtask expressibilityImproved
Design reward with a linear program for the specified taskFinite CMPs with SOAP, PO, or TO tasks · Reward designcomputational complexityImproved
Restrict reward functions to finitely many outputsPO reward design · Finite-PO-RewardDecisioncomputational complexityHurt
Require one reward function to realize a task across multiple environmentsA task realizable in each CMP separately · Sets of CMPs with shared state-action spacetask expressibilityHurt
Use a SOAP-designed reward instead of the regular goal-based rewardQ-learning in the grid world · Grid worldmatch to an acceptable policyImproved
Increase the spread of acceptable policiesSOAP reward expressibility · Small CMPsfraction of expressible SOAPsHurt

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy