NPMixer: Hierarchical Neighboring Patch Mixing for Time Series Forecasting

TL;DR

NPMixer is a hierarchical neural network architecture that adaptively decomposes time series signals into trend and detail components, capturing multi-scale dependencies for improved forecasting accuracy.

Contribution

It introduces a Learnable Stationary Wavelet Transform and hierarchical MLP-based mixer blocks for adaptive, multi-scale time series decomposition and modeling.

Findings

Outperforms state-of-the-art models on 20 out of 28 benchmark setups.

Achieves superior MSE performance across seven datasets.

Effectively captures local and global temporal dependencies.

Abstract

Multivariate time series forecasting remains a challenge due to the complexity of local temporal dynamics and global dependencies across multiple variables. In this paper, we propose \textbf{N}eighboring \textbf{P}atching \textbf{Mixer} (\textbf{NPMixer}), a hierarchical architecture featuring a Learnable Stationary Wavelet Transform that adaptively learns filter coefficients to decompose signals into trend and detail components in a data-dependent manner. Our framework introduces a Neighboring Mixer Block that captures local temporal dynamics through a series of hierarchical MLP layers operating on non-overlapping patches. Specifically, the mixer block utilizes MLPs to learn temporal patterns within and across these patches, expanding the receptive field to capture multi-scale dependencies. A Channel-Mixing Encoder is applied to high-frequency components to learn channel correlations while preserving the stability of the underlying global trend. Extensive experiments on seven benchmark datasets demonstrate that NPMixer consistently outperforms state-of-the-art models, achieving better performance in 20 out of 28 ($71.4\%$) evaluated experimental setups for MSE.