A Predictive Approach To Enhance Time-Series Forecasting

TL;DR

This paper introduces Future-Guided Learning, a dynamic feedback approach inspired by predictive coding, significantly improving long-term dependency capture and adaptability in time-series forecasting tasks.

Contribution

It proposes a novel feedback mechanism with detection and forecasting models that adaptively minimize surprise, enhancing deep learning performance in time-series prediction.

Findings

44.8% increase in AUC-ROC for seizure prediction

23.4% reduction in MSE for nonlinear systems

Effective dynamic adjustment of model parameters

Abstract

Accurate time-series forecasting is crucial in various scientific and industrial domains, yet deep learning models often struggle to capture long-term dependencies and adapt to data distribution shifts over time. We introduce Future-Guided Learning, an approach that enhances time-series event forecasting through a dynamic feedback mechanism inspired by predictive coding. Our method involves two models: a detection model that analyzes future data to identify critical events and a forecasting model that predicts these events based on current data. When discrepancies occur between the forecasting and detection models, a more significant update is applied to the forecasting model, effectively minimizing surprise, allowing the forecasting model to dynamically adjust its parameters. We validate our approach on a variety of tasks, demonstrating a 44.8% increase in AUC-ROC for seizure prediction using EEG data, and a 23.4% reduction in MSE for forecasting in nonlinear dynamical systems (outlier excluded).By incorporating a predictive feedback mechanism, Future-Guided Learning advances how deep learning is applied to time-series forecasting.