SAFE: Spectral Evolution Analysis Feature Extraction for Non-Stationary Time Series Prediction

TL;DR

SAFE is a real-time spectral evolution-based method for detecting non-stationarity in time series, enabling adaptive predictions with reduced computational costs.

Contribution

The paper introduces SAFE, a novel spectral evolution analysis technique that detects non-stationarity and guides online adaptation in machine learning models for time series prediction.

Findings

Significantly reduces computational resource usage.

Maintains high prediction accuracy during non-stationary periods.

Effective on both artificial and real-world datasets.

Abstract

This paper presents a practical approach for detecting non-stationarity in time series prediction. This method is called SAFE and works by monitoring the evolution of the spectral contents of time series through a distance function. This method is designed to work in combination with state-of-the-art machine learning methods in real time by informing the online predictors to perform necessary adaptation when a non-stationarity presents. We also propose an algorithm to proportionally include some past data in the adaption process to overcome the Catastrophic Forgetting problem. To validate our hypothesis and test the effectiveness of our approach, we present comprehensive experiments in different elements of the approach involving artificial and real-world datasets. The experiments show that the proposed method is able to significantly save computational resources in term of processor or GPU cycles while maintaining high prediction performances.