Search papers, labs, and topics across Lattice.
The paper introduces FreqCycle, a time series forecasting framework that integrates time-domain cycle forecasting for low-frequency features with frequency-domain pattern learning for mid-to-high frequency components. To handle coupled multi-periodicity and long lookback windows, a hierarchical extension, MFreqCycle, is proposed to decouple nested periodic features across scales. Experiments on seven benchmarks show FreqCycle achieves state-of-the-art accuracy with faster inference speeds.
By explicitly modeling mid-to-high frequency patterns often ignored by existing methods, FreqCycle unlocks state-of-the-art time series forecasting accuracy while maintaining faster inference.
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.
