We prove that two popular linear contextual bandit algorithms, OFUL and Thompson Sampling, can be made efficient using Frequent Directions, a deterministic online sketching technique. More precisely, we show that a sketch of size m allows a O(md) update time for both algorithms, as opposed to Ω(d 2 ) required by their non-sketched versions in general (where d is the dimension of context vectors). This computational speedup is accompanied by regret bounds of order (1 + εm) 3/2d √ T for OFUL and of order (1 + εm)d 3/2√ T for Thompson Sampling, where εm is bounded by the sum of the tail eigenvalues not covered by the sketch. In particular, when the selected contexts span a subspace of dimension at most m, our algorithms have a regret bound matching that of their slower, non-sketched counterparts. Experiments on real-world datasets corroborate our theoretical results.

Efficient Linear Bandits through Matrix Sketching / I. Kuzborskij, L. Cella, N. Cesa-Bianchi (PROCEEDINGS OF MACHINE LEARNING RESEARCH). - In: International Conference on Artificial Intelligence and Statistics / [a cura di] K. Chaudhuri, M. Sugiyama. - [s.l] : PMLR, 2019. - pp. 177-185 (( Intervento presentato al 22. convegno International Conference on Artificial Intelligence and Statistics nel 2019.

Efficient Linear Bandits through Matrix Sketching

I. Kuzborskij;L. Cella;N. Cesa-Bianchi
2019

Abstract

We prove that two popular linear contextual bandit algorithms, OFUL and Thompson Sampling, can be made efficient using Frequent Directions, a deterministic online sketching technique. More precisely, we show that a sketch of size m allows a O(md) update time for both algorithms, as opposed to Ω(d 2 ) required by their non-sketched versions in general (where d is the dimension of context vectors). This computational speedup is accompanied by regret bounds of order (1 + εm) 3/2d √ T for OFUL and of order (1 + εm)d 3/2√ T for Thompson Sampling, where εm is bounded by the sum of the tail eigenvalues not covered by the sketch. In particular, when the selected contexts span a subspace of dimension at most m, our algorithms have a regret bound matching that of their slower, non-sketched counterparts. Experiments on real-world datasets corroborate our theoretical results.
Settore INF/01 - Informatica
2019
http://proceedings.mlr.press/v89/kuzborskij19a.html
Book Part (author)
File in questo prodotto:
File Dimensione Formato  
kuzborskij19a.pdf

accesso aperto

Tipologia: Publisher's version/PDF
Dimensione 2.01 MB
Formato Adobe PDF
2.01 MB Adobe PDF Visualizza/Apri
Pubblicazioni consigliate

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/2434/640342
Citazioni
  • ???jsp.display-item.citation.pmc??? ND
  • Scopus ND
  • ???jsp.display-item.citation.isi??? 6
social impact