Interpretation of Neural Network Players for a Generalized Divide the Dollar Game Using SHAP Values

Garrison W. Greenwood, Hussin Abbass, Aya Hussein

Research output: A Conference proceeding or a Chapter in BookConference contributionpeer-review

Abstract

Machine learning models can make accurate predictions but trust in the models depends on being able to under-stand why those predictions were made. Unfortunately, machine learning models are black boxes making interpretation difficult. Previously we used an evolutionary algorithm to evolve triplets of neural network players for instances of the Generalized Divide-the-Dollar, which is an economic bargaining game. The players produced fair bids with high bid totals, which is a desirable outcome, but no attempt was made to understand why the players performed so well. In this paper, we interpret the behavior of those neural networks using SHapley Additive exPlanations (or SHAP). Surprisingly, the neural network players exhibited both altruistic and exploitative behavior. Both a global and a local interpretation analysis is presented. The experiments conducted in this work demonstrate a simple method for understanding players' strategies in multi-player gamcs.

Original languageEnglish
Title of host publication2023 IEEE Symposium Series on Computational Intelligence, SSCI 2023
EditorsWen Yu, Agung Julius, Minwoo Jake Lee, Dianhui Wang, Zhi-Hui Zhan, Xiaoou Li
PublisherIEEE, Institute of Electrical and Electronics Engineers
Pages1808-1813
Number of pages6
ISBN (Electronic)9781665430654
DOIs
Publication statusPublished - Dec 2023
Externally publishedYes
Event2023 IEEE Symposium Series on Computational Intelligence, SSCI 2023 - Mexico City, Mexico
Duration: 5 Dec 20238 Dec 2023

Publication series

Name2023 IEEE Symposium Series on Computational Intelligence, SSCI 2023

Conference

Conference2023 IEEE Symposium Series on Computational Intelligence, SSCI 2023
Country/TerritoryMexico
CityMexico City
Period5/12/238/12/23

Fingerprint

Dive into the research topics of 'Interpretation of Neural Network Players for a Generalized Divide the Dollar Game Using SHAP Values'. Together they form a unique fingerprint.

Cite this