RhyMix: A Lightweight Adaptive Multi-Rhythm Network for Long-Term Time Series Forecasting

2026-07-09

AI Summary

Real-world time series exhibit diverse concurrent temporal patterns—including short-term fluctuations, seasonality, long-term trends, and abrupt disturbances—posing challenges for existing models to simultaneously capture local details and global dependencies. This work proposes RhyMix, a lightweight adaptive multi-rhythm network featuring a parallel dual-path architecture: a recurrent path modeling seasonal regularities and a multi-scale temporal convolutional path capturing dependencies across granularities. These paths are dynamically fused via a multi-level adaptive gating mechanism that integrates four specialized forecasting heads. Combining learnable recurrent embeddings, deep dilated convolutions, and channel-wise attention, RhyMix achieves linear computational complexity with only approximately 40K parameters. Evaluated on twelve real-world long-term forecasting benchmarks, it attains state-of-the-art performance on ten datasets, with inference latency under 5ms, demonstrating strong suitability for low-resource and edge deployment scenarios.

Abstract

Real-world time series exhibit complex dynamics characterized by multiple simultaneous temporal patterns: short-term fluctuations, periodic seasonal cycles, long-term trends, and irregular abrupt changes. However, many existing forecasting architectures rely on single-path temporal modeling--transformers capture long-range dependencies but smooth local variations, convolutions capture local patterns but have limited receptive fields, and linear models are efficient but cannot capture nonlinear dynamics. To address this, we introduce RhyMix (RHYthm MIXture), a hybrid neural architecture designed around a parallel dual-path modeling paradigm with adaptive gating mechanisms. RhyMix integrates two complementary encoding branches: (i) a Cyclic Path that incorporates explicit seasonal inductive bias through learnable cyclic embeddings, capturing predictable rhythmic patterns; and (ii) a lightweight Multi-Scale Temporal Convolutional Network with Channel Attention Path that employs multi-scale depthwise dilated convolutions to capture temporal dependencies across different receptive fields. A key innovation is the use of adaptive gating at multiple levels: a path gate dynamically combines four specialized forecasting heads (Direct, Trend-Seasonal Decomposition, Local Convolution, and Periodic Fusion) per sample and channel, while a hybrid gate adaptively balances the Cyclic and MSTCN-CA Paths based on input characteristics. This design ensures the model adapts to specific temporal patterns while maintaining linear complexity in sequence length, channels, and prediction horizon. Across extensive benchmarks on 12 real-world datasets for long-term forecasting, RhyMix achieves state-of-the-art performance on 10 of 12 datasets. The model remains lightweight (~40K params) with linear complexity and low-latency inference (<5ms),suitable for resource-constrained edge devices and real-time deployment.

Problem

• time series forecasting
• multi-rhythm modeling
• long-term prediction
• temporal patterns
• adaptive modeling

Innovation

• adaptive gating
• multi-rhythm modeling
• lightweight architecture
• dual-path network
• long-term time series forecasting