ScatterFusion: A Hierarchical Scattering Transform Framework for Enhanced Time Series Forecasting

TL;DR

ScatterFusion is a new hierarchical framework combining scattering transforms and attention mechanisms to improve time series forecasting by capturing multi-scale patterns and dependencies, leading to better accuracy.

Contribution

It introduces a novel hierarchical scattering transform framework with scale-adaptive features and multi-resolution attention for enhanced forecasting accuracy.

Findings

Outperforms existing methods on seven benchmark datasets.

Achieves significant error reduction across multiple prediction horizons.

Effectively captures multi-scale temporal dependencies.

Abstract

Time series forecasting presents significant challenges due to the complex temporal dependencies at multiple time scales. This paper introduces ScatterFusion, a novel framework that synergistically integrates scattering transforms with hierarchical attention mechanisms for robust time series forecasting. Our approach comprises four key components: (1) a Hierarchical Scattering Transform Module (HSTM) that extracts multi-scale invariant features capturing both local and global patterns; (2) a Scale-Adaptive Feature Enhancement (SAFE) module that dynamically adjusts feature importance across different scales; (3) a Multi-Resolution Temporal Attention (MRTA) mechanism that learns dependencies at varying time horizons; and (4) a Trend-Seasonal-Residual (TSR) decomposition-guided structure-aware loss function. Extensive experiments on seven benchmark datasets demonstrate that ScatterFusion outperforms other common methods, achieving significant reductions in error metrics across various prediction horizons.