Universality of Bayesian mixture predictors - Archive ouverte HAL Access content directly
Conference Papers Year :

Universality of Bayesian mixture predictors

(1)
1
Daniil Ryabko
  • Function : Author
  • PersonId : 848126

Abstract

The problem is that of sequential probability forecasting for finite-valued time series. The data is generated by an unknown probability distribution over the space of all one-way infinite sequences. It is known that this measure belongs to a given set C, but the latter is completely arbitrary (uncountably infinite, without any structure given). The performance is measured with asymptotic average log loss. In this work it is shown that the minimax asymptotic performance is always attainable, and it is attained by a convex combination of a countably many measures from the set C (a Bayesian mixture). This was previously only known for the case when the best achievable asymptotic error is 0. This also contrasts previous results that show that in the non-realizable case all Bayesian mixtures may be suboptimal, while there is a predictor that achieves the optimal performance.

Dates and versions

hal-01627332 , version 1 (01-11-2017)

Identifiers

Cite

Daniil Ryabko. Universality of Bayesian mixture predictors. ALT 2017 - 28th International Conference on Algorithmic Learning Theory, Oct 2017, Kyoto, Japan. pp.1-13. ⟨hal-01627332⟩
57 View
0 Download

Altmetric

Share

Gmail Facebook Twitter LinkedIn More