Evaluating Agents using Social Choice Theory

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Lanctot, Marc, Larson, Kate, Bachrach, Yoram, Marris, Luke, Li, Zun, Bhoopchand, Avishkar, Anthony, Thomas, Tanner, Brian, Koop, Anna
Format: Preprint
Published: 2023
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866918073845415936
author Lanctot, Marc
Larson, Kate
Bachrach, Yoram
Marris, Luke
Li, Zun
Bhoopchand, Avishkar
Anthony, Thomas
Tanner, Brian
Koop, Anna
author_facet Lanctot, Marc
Larson, Kate
Bachrach, Yoram
Marris, Luke
Li, Zun
Bhoopchand, Avishkar
Anthony, Thomas
Tanner, Brian
Koop, Anna
contents We argue that many general evaluation problems can be viewed through the lens of voting theory. Each task is interpreted as a separate voter, which requires only ordinal rankings or pairwise comparisons of agents to produce an overall evaluation. By viewing the aggregator as a social welfare function, we are able to leverage centuries of research in social choice theory to derive principled evaluation frameworks with axiomatic foundations. These evaluations are interpretable and flexible, while avoiding many of the problems currently facing cross-task evaluation. We apply this Voting-as-Evaluation (VasE) framework across multiple settings, including reinforcement learning, large language models, and humans. In practice, we observe that VasE can be more robust than popular evaluation frameworks (Elo and Nash averaging), discovers properties in the evaluation data not evident from scores alone, and can predict outcomes better than Elo in a complex seven-player game. We identify one particular approach, maximal lotteries, that satisfies important consistency properties relevant to evaluation, is computationally efficient (polynomial in the size of the evaluation data), and identifies game-theoretic cycles.
format Preprint
id arxiv_https___arxiv_org_abs_2312_03121
institution arXiv
publishDate 2023
record_format arxiv
spellingShingle Evaluating Agents using Social Choice Theory
Lanctot, Marc
Larson, Kate
Bachrach, Yoram
Marris, Luke
Li, Zun
Bhoopchand, Avishkar
Anthony, Thomas
Tanner, Brian
Koop, Anna
Artificial Intelligence
Computer Science and Game Theory
Multiagent Systems
We argue that many general evaluation problems can be viewed through the lens of voting theory. Each task is interpreted as a separate voter, which requires only ordinal rankings or pairwise comparisons of agents to produce an overall evaluation. By viewing the aggregator as a social welfare function, we are able to leverage centuries of research in social choice theory to derive principled evaluation frameworks with axiomatic foundations. These evaluations are interpretable and flexible, while avoiding many of the problems currently facing cross-task evaluation. We apply this Voting-as-Evaluation (VasE) framework across multiple settings, including reinforcement learning, large language models, and humans. In practice, we observe that VasE can be more robust than popular evaluation frameworks (Elo and Nash averaging), discovers properties in the evaluation data not evident from scores alone, and can predict outcomes better than Elo in a complex seven-player game. We identify one particular approach, maximal lotteries, that satisfies important consistency properties relevant to evaluation, is computationally efficient (polynomial in the size of the evaluation data), and identifies game-theoretic cycles.
title Evaluating Agents using Social Choice Theory
topic Artificial Intelligence
Computer Science and Game Theory
Multiagent Systems
url https://arxiv.org/abs/2312.03121