Second-Order Kernel Online Convex Optimization with Adaptive Sketching

Daniele Calandriello 1 Alessandro Lazaric 1 Michal Valko 1
1 SEQUEL - Sequential Learning
Inria Lille - Nord Europe, CRIStAL - Centre de Recherche en Informatique, Signal et Automatique de Lille (CRIStAL) - UMR 9189
Abstract : Kernel online convex optimization (KOCO) is a framework combining the expressiveness of non-parametric kernel models with the regret guarantees of online learning. First-order KOCO methods such as functional gradient descent require only $O(t)$ time and space per iteration, and, when the only information on the losses is their convexity, achieve a minimax optimal $O(\sqrt{T})$ regret. Nonetheless, many common losses in kernel problems, such as squared loss, logistic loss, and squared hinge loss posses stronger curvature that can be exploited. In this case, second-order KOCO methods achieve $O(\log(\Det(K)))$ regret, which we show scales as $O(deff \log T)$, where $deff$ is the effective dimension of the problem and is usually much smaller than $O(\sqrt{T})$. The main drawback of second-order methods is their much higher $O(t^2)$ space and time complexity. In this paper, we introduce kernel online Newton step (KONS), a new second-order KOCO method that also achieves $O(deff\log T)$ regret. To address the computational complexity of second-order methods, we introduce a new matrix sketching algorithm for the kernel matrix~$K$, and show that for a chosen parameter $\gamma \leq 1$ our Sketched-KONS reduces the space and time complexity by a factor of $\gamma^2$ to $O(t^2\gamma^2)$ space and time per iteration, while incurring only $1/\gamma$ times more regret.
Document type :
Conference papers
Complete list of metadatas

Cited literature [27 references]  Display  Hide  Download

https://hal.inria.fr/hal-01537799
Contributor : Michal Valko <>
Submitted on : Tuesday, June 13, 2017 - 3:04:48 AM
Last modification on : Friday, March 22, 2019 - 1:34:50 AM
Long-term archiving on : Tuesday, December 12, 2017 - 12:03:21 PM

File

calandriello2017second-order.p...
Files produced by the author(s)

Identifiers

  • HAL Id : hal-01537799, version 1

Citation

Daniele Calandriello, Alessandro Lazaric, Michal Valko. Second-Order Kernel Online Convex Optimization with Adaptive Sketching. International Conference on Machine Learning, 2017, Sydney, Australia. ⟨hal-01537799⟩

Share

Metrics

Record views

249

Files downloads

124