Streaming kernel regression with provably adaptive mean, variance, and regularization

Audrey Durand 1 Odalric-Ambrym Maillard 2 Joelle Pineau 3
2 SEQUEL - Sequential Learning
Inria Lille - Nord Europe, CRIStAL - Centre de Recherche en Informatique, Signal et Automatique de Lille (CRIStAL) - UMR 9189
Abstract : We consider the problem of streaming kernel regression, when the observations arrive sequentially and the goal is to recover the underlying mean function, assumed to belong to an RKHS. The variance of the noise is not assumed to be known. In this context, we tackle the problem of tuning the regularization parameter adaptively at each time step, while maintaining tight confidence bounds estimates on the value of the mean function at each point. To this end, we first generalize existing results for finite-dimensional linear regression with fixed regularization and known variance to the kernel setup with a regularization parameter allowed to be a measurable function of past observations. Then, using appropriate self-normalized inequalities we build upper and lower bound estimates for the variance, leading to Bersntein-like concentration bounds. The later is used in order to define the adaptive regularization. The bounds resulting from our technique are valid uniformly over all observation points and all time steps, and are compared against the literature with numerical experiments. Finally, the potential of these tools is illustrated by an application to kernelized bandits, where we revisit the Kernel UCB and Kernel Thompson Sampling procedures, and show the benefits of the novel adaptive kernel tuning strategy.
Complete list of metadatas

Cited literature [4 references]  Display  Hide  Download

https://hal.archives-ouvertes.fr/hal-01927007
Contributor : Odalric-Ambrym Maillard <>
Submitted on : Monday, November 19, 2018 - 3:33:00 PM
Last modification on : Thursday, June 27, 2019 - 1:36:06 PM
Long-term archiving on : Wednesday, February 20, 2019 - 3:15:54 PM

File

1708.00768.pdf
Files produced by the author(s)

Identifiers

  • HAL Id : hal-01927007, version 1
  • ARXIV : 1708.00768

Citation

Audrey Durand, Odalric-Ambrym Maillard, Joelle Pineau. Streaming kernel regression with provably adaptive mean, variance, and regularization. Journal of Machine Learning Research, Microtome Publishing, 2018, 1, pp.1 - 48. ⟨hal-01927007⟩

Share

Metrics

Record views

49

Files downloads

38