FreqCycle: A Multi-Scale Time-Frequency Analysis Method for Time Series Forecasting
This work addresses performance gains in time series forecasting for domains like finance or weather, though it appears incremental as it builds on existing frequency analysis methods.
The paper tackled the problem of time series forecasting by addressing the limitation of existing methods that overlook mid to high frequency patterns, proposing FreqCycle to integrate low-frequency and mid to high frequency feature extraction, achieving state-of-the-art accuracy on seven benchmarks with faster inference speeds.
Mining time-frequency features is critical for time series forecasting. Existing research has predominantly focused on modeling low-frequency patterns, where most time series energy is concentrated. The overlooking of mid to high frequency continues to limit further performance gains in deep learning models. We propose FreqCycle, a novel framework integrating: (i) a Filter-Enhanced Cycle Forecasting (FECF) module to extract low-frequency features by explicitly learning shared periodic patterns in the time domain, and (ii) a Segmented Frequency-domain Pattern Learning (SFPL) module to enhance mid to high frequency energy proportion via learnable filters and adaptive weighting. Furthermore, time series data often exhibit coupled multi-periodicity, such as intertwined weekly and daily cycles. To address coupled multi-periodicity as well as long lookback window challenges, we extend FreqCycle hierarchically into MFreqCycle, which decouples nested periodic features through cross-scale interactions. Extensive experiments on seven diverse domain benchmarks demonstrate that FreqCycle achieves state-of-the-art accuracy while maintaining faster inference speeds, striking an optimal balance between performance and efficiency.