Sampling from a Neural Network

Julien Gheysens (), Xiaodan Fan (), Chaojie Wang () and Nicolas Wicker ()
Additional contact information
Julien Gheysens: University of Lille
Xiaodan Fan: The Chinese University of Hong Kong
Chaojie Wang: Jiangsu University
Nicolas Wicker: University of Lille

Sankhya B: The Indian Journal of Statistics, 2024, vol. 86, issue 2, No 20, 872-884

Abstract: Abstract In this article, the video game Dota 2 is used as an illustration of video games where two teams have to confront. This is a very popular game where billions of matches are available for enhancing statistical analysis and fostering further method developments. Here, we focus on sampling teams or pairs of teams favouring more imbalanced matches, based upon the scores observed either for one team or for a couple of teams. An artificial neural network models the expected outcome which is then used to sample teams through a Metropolis-Hastings algorithm. The main result is establishing the polynomial convergence speed bound for both cases, that is sampling random teams and sampling random pairs of teams. An empirical experiment is also provided when sampling pairs of teams for Dota after learning matches outcomes from a database of 2 million matches, showing that the convergence may happen much more quickly.

Keywords: Dota2; team sampling; neural networks; Metropolis-Hastings; markov chain; mixing time (search for similar items in EconPapers)
Date: 2024
References: View complete reference list from CitEc
Citations:

Downloads: (external link)
http://link.springer.com/10.1007/s13571-024-00329-9 Abstract (text/html)
Access to the full text of the articles in this series is restricted.

Related works:
This item may be available elsewhere in EconPapers: Search for items with the same title.

Export reference: BibTeX RIS (EndNote, ProCite, RefMan) HTML/Text

Persistent link: https://EconPapers.repec.org/RePEc:spr:sankhb:v:86:y:2024:i:2:d:10.1007_s13571-024-00329-9

Ordering information: This journal article can be ordered from
http://www.springer.com/statistics/journal/13571

DOI: 10.1007/s13571-024-00329-9

Access Statistics for this article

Sankhya B: The Indian Journal of Statistics is currently edited by Dipak Dey

More articles in Sankhya B: The Indian Journal of Statistics from Springer, Indian Statistical Institute
Bibliographic data for series maintained by Sonal Shukla () and Springer Nature Abstracting and Indexing ().