Continuous Evolution Pool: Taming Recurring Concept Drift in Online Time Series Forecasting

Tianxiang Zhan, Ming Jin, Yuanpeng He, Yuxuan Liang, Yong Deng, Shirui Pan

Keywords: Online Forecasting, Recurring Concepts, Time Series, Continual Learning

TL;DR: To address recurring concept drift in time series, where models forget past patterns and degrade online prediction accuracy, we propose the Continuous Evolution Pool.

Abstract: Recurring concept drift, a type of concept drift in which previously observed data patterns reappear after some time, is one of the most prevalent types of concept drift in time series. As time progresses, concept drift occurs, and previously encountered concepts are forgotten, thereby leading to a decline in the accuracy of online predictions. Existing solutions mainly employ parameter updating techniques to delay forgetting; however, this may result in the loss of some previously learned knowledge while neglecting the exploration of knowledge retention mechanisms. To retain all knowledge and fully utilize it when the concepts recur, we propose the **C**ontinuous **E**volution **P**ool (CEP), a pooling mechanism that stores different instances of forecasters for different concepts. Our method first selects the forecaster nearest to the test sample and then learns the features from its neighboring samples—a process we refer to as **retrieval**. If there are insufficient neighboring samples, it indicates that a new concept has emerged, and a new model will evolve from the current nearest sample to the pool to **store** the knowledge of the concept. Simultaneously, the **elimination** mechanism will enable outdated knowledge to be cleared to ensure the prediction effect of the forecasters. Experiments on real-world datasets demonstrate that by retaining distinct conceptual knowledge, CEP significantly enhances online forecasting accuracy, reducing the error by over 20%.

Primary Area: learning on time series and dynamical systems

Submission Number: 3724