Prediction of Locally Stationary Data Using Expert Advice
This work addresses forecasting for data streams with changing parameters, but it appears incremental as it builds on existing game-theoretic methods without specifying broad impact.
The paper tackles the problem of continuous machine learning for locally stationary time series using a game-theoretic approach without stochastic assumptions, presenting an online forecasting algorithm and obtaining an efficiency estimate.
The problem of continuous machine learning is studied. Within the framework of the game-theoretic approach, when for calculating the next forecast, no assumptions about the stochastic nature of the source that generates the data flow are used -- the source can be analog, algorithmic or probabilistic, its parameters can change at random times, when building a prognostic model, only structural assumptions are used about the nature of data generation. An online forecasting algorithm for a locally stationary time series is presented. An estimate of the efficiency of the proposed algorithm is obtained.