Unimodal Mono-Partite Matching in a Bandit Setting
Résumé
We tackle a new emerging problem, which is finding an optimal monopartite matching in a weighted graph.
The semi-bandit version, where a full matching is sampled at each iteration, has been addressed by \cite{ADMA}, creating an algorithm with an expected regret matching $O(\frac{L\log(L)}{\Delta}\log(T))$ with $2L$ players, $T$ iterations and a minimum reward gap $\Delta$.
We reduce this bound in two steps. First, as in \cite{GRAB} and \cite{UniRank} we use the unimodality property of the expected reward on the appropriate graph to design an algorithm with a regret in $O(L\frac{1}{\Delta}\log(T))$.
Secondly, we show that by moving the focus towards the main question `\emph{Is user $i$ better than user $j$?}' this regret becomes $O(L\frac{\Delta}{\tilde{\Delta}^2}\log(T))$, where $\Tilde{\Delta} > \Delta$ derives from a better way of comparing users.
Some experimental results finally show these theoretical results are corroborated in practice.
Domaines
Apprentissage [cs.LG]
Fichier principal
main.pdf (356.13 Ko)
Télécharger le fichier
main.synctex.gz (157.06 Ko)
Télécharger le fichier
Origine | Fichiers produits par l'(les) auteur(s) |
---|