SDMixer: Sparse Dual-Mixer for Time Series Forecasting

Xiang Ao

cs.LGFeb 27, 2026

SDMixer: Sparse Dual-Mixer for Time Series Forecasting

Authors: Xiang Ao

Organizations: School of Software Engineering, Beijing Jiaotong University, Beijing, China

Abstract

Multivariate time series forecasting is widely applied in fields such as transportation, energy, and finance. However, the data commonly suffers from issues of multi-scale characteristics, weak correlations, and noise interference, which limit the predictive performance of existing models. This paper proposes a dual-stream sparse Mixer prediction framework that extracts global trends and local dynamic features from sequences in both the frequency and time domains, respectively. It employs a sparsity mechanism to filter out invalid information, thereby enhancing the accuracy of cross-variable dependency modeling. Experimental results demonstrate that this method achieves leading performance on multiple real-world scenario datasets, validating its effectiveness and generality. The code is available at https://github.com/SDMixer/SDMixer

Figures & tables

Dataset	Dim	Dataset Size	Forecastability*	Information	$\frac{\text{cov(Season)}}{\text{cov(Trend)}}$
ETTm1	7	$(34465,11521,11521)$	0.46	Temperature	0.073992
ETTm2	7	$(34465,11521,11521)$	0.55	Temperature	0.039469
ETTh1	7	$(8545,2881,2881)$	0.38	Temperature	0.490909
ETTh2	7	$(8545,2881,2881)$	0.45	Temperature	0.136543
Electricity	321	$(18317,2633,5261)$	0.77	Electricity	11.836820
Exchange	8	$(5312,760,1516)$	0.42	Economy	0.001885

Table 1: Dataset detailed descriptions. The dataset size is organized in (Train, Validation, Test). Forecastability is calculated by one minus the entropy of Fourier decomposition of time series [ 4 ] ; a larger value indicates better predictability.


		SDMixer		WPMixer		TimeMixer		iTransformer		PatchTST
		MSE	MAE	MSE	MAE	MSE	MAE	MSE	MAE	MSE	MAE

Weather	96	0.168	0.221	0.163	0.205	0.163	0.209	0.174	0.214	0.186	0.227
	192	0.210	0.256	0.208	0.245	0.208	0.250	0.221	0.254	0.234	0.265
	336	0.256	0.291	0.263	0.287	0.259	0.287	0.278	0.296	0.284	0.301

Table 2: Comparative experimental results with typical baselines. The best performance is in bold.

Dataset	SDMixer		TimesNet		DLinear		SCINet		FEDformer		Autoformer
Dataset	MSE	MAE	MSE	MAE	MSE	MAE	MSE	MAE	MSE	MAE	MSE	MAE
Weather	0.238	0.276	0.259	0.287	0.265	0.315	0.292	0.363	0.309	0.360	0.338	0.382
Electricity	0.167	0.262	0.192	0.304	0.225	0.319	0.268	0.365	0.214	0.327	0.227	0.338
Exchange	0.355	0.399	0.416	0.443	0.354	0.414	0.750	0.626	0.519	0.429	0.613	0.539
ETTh1	0.426	0.438	0.458	0.450	0.461	0.457	0.747	0.647	0.498	0.484	0.496	0.487
ETTh2	0.343	0.391	0.414	0.427	0.563	0.519	0.954	0.723	0.437	0.449	0.450	0.459

Table 3: Comparative experimental results with more baselines. For each dataset, the result is the average over the four prediction lengths (96, 192, 336, 720). The best performance is in bold.

Model Variant	Metric	Datasets							AVG
Model Variant	Metric	ECL	ETTh1	ETTh2	ETTm1	ETTm2	Exchange	Weather	AVG
No Cross	MSE	0.170	0.436	0.364	0.381	0.278	0.375	0.279	0.326
No Cross	MAE	0.271	0.448	0.421	0.384	0.359	0.429	0.338	0.379
No Time	MSE	0.178	0.424	0.344	0.356	0.268	0.408	0.249	0.318
No Time	MAE	0.279	0.440	0.395	0.383	0.329	0.439	0.287	0.365
No Freq	MSE	0.193	0.469	0.367	0.374	0.258	0.355	0.239	0.322

Table 4: Ablation experiment results. For each dataset, the result is the average over the four prediction lengths (96, 192, 336, 720). The best performance is in bold.

Figure 2: Performance of ablation models on different datasets (errors for prediction length 96).

Explore similar work

NPMixer: Hierarchical Neighboring Patch Mixing for Time Series Forecasting
May 8, 2026Jung Min Choi, Vijaya Krishna Yalavarthi, Lars Schmidt-ThiemeMultivariate Time Series Forecasting Time Series Forecasting
1ISMLL, University of Hildesheim, VWFS Data Analytics Research Center (VWFS-DARC), Hildesheim, Germany.
DualWeaver: Synergistic Feature Weaving Surrogates for Multivariate Forecasting with Univariate Time Series Foundation Models
Feb 25, 2026Jinpeng Li, Zhongyi Pei, Huaze Xue +3Multivariate Time Series Forecasting Time Series Forecasting
School of Software, BNRist, Tsinghua University · Tencent
SEMixer: Semantics Enhanced MLP-Mixer for Multiscale Mixing and Long-term Time Series Forecasting
Feb 18, 2026Xu Zhang, Qitong Wang, Peng Wang +1Long-Term Time Series Forecasting Multi-Scale Feature Fusion
Shanghai Key Laboratory of Data Science, College of Computer Science and Artificial Intelligence, Fudan University, Shanghai, China · Harvard University, Cambridge, Massachusetts, United States

May 8, 2026J/K move · Enter open · S save

NPMixer: Hierarchical Neighboring Patch Mixing for Time Series Forecasting

Jung Min Choi, Vijaya Krishna Yalavarthi, Lars Schmidt-Thieme

1ISMLL, University of Hildesheim, VWFS Data Analytics Research Center (VWFS-DARC), Hildesheim, Germany.

Multivariate Time Series Forecasting Time Series Forecasting Wavelet Methods Time Series Decomposition Multiscale Time-Series Modeling

Open paper AskPDF arXiv

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 Neighboring Patching Mixer (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.

Abstract

Explore similar work

NPMixer: Hierarchical Neighboring Patch Mixing for Time Series Forecasting

DualWeaver: Synergistic Feature Weaving Surrogates for Multivariate Forecasting with Univariate Time Series Foundation Models

SEMixer: Semantics Enhanced MLP-Mixer for Multiscale Mixing and Long-term Time Series Forecasting