Predicting future dynamics from short-term time series using an Anticipated Learning Machine. Issue 6 (19th February 2020)

• Record Type: Journal Article
• Authors: Chen, Chuan; Li, Rui; Shu, Lin; He, Zhiyu; Wang, Jining; Zhang, Chengming; Ma, Huanfei; Aihara, Kazuyuki; Chen, Luonan
• Is Part Of: National science review. Volume 7:Issue 6(2020)
• Page Start: 1079
• Page End: 1091
• Publication Date: 2020-02-19
• Subjects: dynamics-based machine learning -- delay embedding theory -- short-term time series prediction -- dynamics-based data science; Science -- Periodicals; 505
• DOI: 10.1093/nsr/nwaa025
• Languages: English
• ISSNs: 2095-5138

Abstract

Abstract: Predicting time series has significant practical applications over different disciplines. Here, we propose an Anticipated Learning Machine (ALM) to achieve precise future-state predictions based on short-term but high-dimensional data. From non-linear dynamical systems theory, we show that ALM can transform recent correlation/spatial information of high-dimensional variables into future dynamical/temporal information of any target variable, thereby overcoming the small-sample problem and achieving multistep-ahead predictions. Since the training samples generated from high-dimensional data also include information of the unknown future values of the target variable, it is called anticipated learning. Extensive experiments on real-world data demonstrate significantly superior performances of ALM over all of the existing 12 methods. In contrast to traditional statistics-based machine learning, ALM is based on non-linear dynamics, thus opening a new way for dynamics-based machine learning. Abstract : We propose Anticipated Learning Machine, a dynamics-based data-driven method, to achieve future-state predictions based on short-term but high-dimensional data, which transforms recent spatial information of high-dimensional variables into future temporal information of any target variable.