MMFNet: Multi-Scale Frequency Masking Neural Network for Multivariate Time Series Forecasting

TL;DR

MMFNet is a novel neural network that uses multi-scale frequency masking to improve long-term multivariate time series forecasting by capturing diverse temporal patterns and filtering irrelevant frequency components.

Contribution

The paper introduces MMFNet, a new model that employs learnable frequency masks at multiple scales to enhance long-term forecasting accuracy.

Findings

Achieves up to 6.0% reduction in MSE compared to state-of-the-art models.

Effectively captures multi-scale temporal patterns in multivariate time series.

Demonstrates robustness across various benchmark datasets.

Abstract

Long-term Time Series Forecasting (LTSF) is critical for numerous real-world applications, such as electricity consumption planning, financial forecasting, and disease propagation analysis. LTSF requires capturing long-range dependencies between inputs and outputs, which poses significant challenges due to complex temporal dynamics and high computational demands. While linear models reduce model complexity by employing frequency domain decomposition, current approaches often assume stationarity and filter out high-frequency components that may contain crucial short-term fluctuations. In this paper, we introduce MMFNet, a novel model designed to enhance long-term multivariate forecasting by leveraging a multi-scale masked frequency decomposition approach. MMFNet captures fine, intermediate, and coarse-grained temporal patterns by converting time series into frequency segments at varying scales while employing a learnable mask to filter out irrelevant components adaptively. Extensive experimentation with benchmark datasets shows that MMFNet not only addresses the limitations of the existing methods but also consistently achieves good performance. Specifically, MMFNet achieves up to 6.0% reductions in the Mean Squared Error (MSE) compared to state-of-the-art models designed for multivariate forecasting tasks.