Learning Theory and Algorithms for Forecasting Non-stationary Time Series

Part of Advances in Neural Information Processing Systems 28 (NIPS 2015)

Authors

Vitaly Kuznetsov, Mehryar Mohri

Abstract

We present data-dependent learning bounds for the general scenario of non-stationary non-mixing stochastic processes. Our learning guarantees are expressed in terms of a data-dependent measure of sequential complexity and a discrepancy measure that can be estimated from data under some mild assumptions. We use our learning bounds to devise new algorithms for non-stationary time series forecasting for which we report some preliminary experimental results.