Online Learning for Time Series Prediction
arXiv:1302.6927
Abstract
In this paper we address the problem of predicting a time series using the ARMA (autoregressive moving average) model, under minimal assumptions on the noise terms. Using regret minimization techniques, we develop effective online learning algorithms for the prediction problem, without assuming that the noise terms are Gaussian, identically distributed or even independent. Furthermore, we show that our algorithm's performances asymptotically approaches the performance of the best ARMA model in hindsight.
17 pages, 6 figures
Cited by in corpus (18)
- Robust Sequential Change-Point Detection by Convex Optimization
- Spectral Filtering for General Linear Dynamical Systems
- Aggregation of predictors for nonstationary sub-linear processes and online adaptive forecasting of time varying autoregressive processes
- Fairness in Forecasting and Learning Linear Dynamical Systems
- No-Regret Prediction in Marginally Stable Systems
- Autoregressive-Model-Based Methods for Online Time Series Prediction with Missing Values: an Experimental Evaluation
- Robust guarantees for learning an autoregressive filter
- An Iterative Riccati Algorithm for Online Linear Quadratic Control
- Improved rates for prediction and identification of partially observed linear dynamical systems
- Nearly second-order asymptotic optimality of sequential change-point detection with one-sample updates
- NonSTOP: A NonSTationary Online Prediction Method for Time Series
- Personalized Online Machine Learning
- POLA: Online Time Series Prediction by Adaptive Learning Rates
- Dynamic Online Gradient Descent with Improved Query Complexity: A Theoretical Revisit
- Online Learning Algorithms for Quaternion ARMA Model
- Worst-case Prediction Performance Analysis of the Kalman Filter
- Online Estimation for Functional Data
- Online Convex Optimization in Changing Environments and its Application to Resource Allocation