MR-ImagenTime: Multi-Resolution Time Series Generation through Dual Image Representations

Xianyong Xu; Yuanjun Zuo; Zhihong Huang; Yihan Qin; Haoxian Xu; Leilei Du; Haotian Wang

cs.LGMar 30, 2026

MR-ImagenTime: Multi-Resolution Time Series Generation through Dual Image Representations

Authors: Xianyong Xu, Yuanjun Zuo, Zhihong Huang, Yihan Qin, Haoxian Xu, Leilei Du, Haotian Wang

Organizations: Department, State Grid Hunan Electric Power Company Limited Research Institute & Hunan Province Engineering Technology Research Center of Electric Power Multimodal Perception and Edge Intelligence, Changsha, Hunan, China. · College of Computer Science, Hunan University, Changsha, Hunan, China.

Abstract

Time series forecasting is vital across many domains, yet existing models struggle with fixed-length inputs and inadequate multi-scale modeling. We propose MR-CDM, a framework combining hierarchical multi-resolution trend decomposition, an adaptive embedding mechanism for variable-length inputs, and a multi-scale conditional diffusion process. Evaluations on four real-world datasets demonstrate that MR-CDM significantly outperforms state-of-the-art baselines (e.g., CSDI, Informer), reducing MAE and RMSE by approximately 6-10 to a certain degree.

Figures & tables

Model	MSE	MAE	RMSE
ARIMA	42.2121	6.2893	6.496
LSTM	13.4981	3.6270	3.6739
CSDI	1.5538	0.8992	1.2465
MR-CDM	1.4842	0.9650	1.2183

Table 1: Performance comparison on ETTh1.

Figure 2: Multi-Step Prediction Performance Comparison.

Appendix figures & tables12 assets

Supplementary material from the paper’s appendix.

Appendix

Model-Variant	MSE	MAE	RMSE
Baseline-NoDecomposition	14.2177	3.6753	3.7706
UnconditionalDiffusion	20.4822	4.4252	4.5257
NoImageFusion	13.0950	3.5406	3.6187
FullModel	1.4842	0.9650	1.2183

Appendix

Table 2: Ablation Studies on ETTh1.

Figure 3: Prediction Result Based on MR- CDM model without conditional information.

Model-Variant	MSE	MAE	RMSE
Baseline-NoDecomposition	108.5863	8.0619	10.4204
UnconditionalDiffusion	116.4437	8.3155	10.7909
NoImageFusion	111.8180	8.2349	10.5744
FullModel	1.2544	0.9080	1.1200

Appendix

Table 3: Ablation Studies on Synthetic ETTh1.

Model-Variant	MSE	MAE	RMSE
MR-CDM w/o Trend1	1.9342	1.1247	1.3907
MR-CDM w/o Trend3	2.2769	1.2678	1.5089
FullModel	1.4842	0.9650	1.2183

Appendix

Table 4: Trend Decomposition Ablation Result.

Seq_Len	MSE	MAE	RMSE
48	1.8998	1.1387	1.3783
96	1.4842	0.9650	1.2183
192	1.2319	0.8492	1.0965

Appendix

Table 5: Input Length Sensitivity Results.

Model	MSE	MAE	RMSE
ARIMA	191.2356	11.6258	13.8287
LSTM	124.7143	7.5514	11.1675
CSDI	96.3245	6.9438	9.8145
MR-CDM	79.4128	6.3487	8.9113

Appendix

Table 6: Model Performance on Weather Domain.

Model	MSE	MAE	RMSE
ARIMA	0.0011	0.0237	0.0331
LSTM	0.0007	0.0152	0.0264
CSDI	0.0004	0.0143	0.0200
MR-CDM	0.0003	0.0139	0.0173

Appendix

Table 7: Model Performance on Traffic Domain.

Model	MSE	MAE	RMSE
ARIMA	30.0946	8.2296	5.4858
LSTM	13.2883	6.2361	3.6453
CSDI	2.7493	1.2478	1.6581

Appendix

Table 8: Multiple-Run Performance on ETTh1.

Model	MSE	MAE	RMSE
ARIMA	47.2826	7.4434	6.8762
LSTM	13.6146	3.6670	3.6897
CSDI	2.6159	1.9632	1.6173

Appendix

Table 9: Multiple-Run Performance on Synsetic ETTh1.

Model	MSE	MAE	RMSE
ARIMA	39.8131	8.3046	6.3097
LSTM	12.7953	8.2361	3.5770
CSDI	1.6749	0.8741	1.2941
MR-CDM	1.2544	0.9080	1.1200

Appendix

Table 10: Performance comparison on Sythetic Dataset.

Figure 4: The Prediction Result Based on CSDI Model.

Figure 5: The Prediction Result Based on MR-CDM Model.

Explore similar work

DMSC: Dynamic Multi-Scale Coordination Framework for Time Series Forecasting
Aug 3, 2025Haonan Yang, Jianchao Tang, Zhuo Li +1Time Series Forecasting Multi-Scale Feature Fusion
National University of Defense Technology, Changsha, Hunan, China
Generalizing Multi-Scale Time-Series Modeling with a Single Operator
May 29, 2026Cheonwoo Lee, Dooho Lee, Doyun Choi +1Long-Term Time Series Forecasting Time Series Forecasting
School of Electrical Engineering, KAIST, Daejeon, Republic of Korea · Department of Computer Science and Engineering, Seoul National University, Seoul, Republic of Korea
SDMixer: Sparse Dual-Mixer for Time Series Forecasting
Feb 27, 2026Xiang AoMultivariate Time Series Forecasting Time Series Forecasting
School of Software Engineering, Beijing Jiaotong University, Beijing, China

Aug 3, 2025J/K move · Enter open · S save

DMSC: Dynamic Multi-Scale Coordination Framework for Time Series Forecasting

Haonan Yang, Jianchao Tang, Zhuo Li, Long Lan

National University of Defense Technology, Changsha, Hunan, China

Time Series Forecasting Multi-Scale Feature Fusion Time Series Decomposition Multiscale Time-Series Modeling

Open paper AskPDF arXiv

Time Series Forecasting (TSF) faces persistent challenges in modeling intricate temporal dependencies across different scales. Despite recent advances leveraging different decomposition operations and novel architectures based on CNN, MLP or Transformer, existing methods still struggle with static decomposition strategies, fragmented dependency modeling, and inflexible fusion mechanisms, limiting their ability to model intricate temporal dependencies. To explicitly solve the mentioned three problems respectively, we propose a novel Dynamic Multi-Scale Coordination Framework (DMSC) with Multi-Scale Patch Decomposition block (EMPD), Triad Interaction Block (TIB) and Adaptive Scale Routing MoE block (ASR-MoE). Specifically, EMPD is designed as a built-in component to dynamically segment sequences into hierarchical patches with exponentially scaled granularities, eliminating predefined scale constraints through input-adaptive patch adjustment. TIB then jointly models intra-patch, inter-patch, and cross-variable dependencies within each layer's decomposed representations. EMPD and TIB are jointly integrated into layers forming a multi-layer progressive cascade architecture, where coarse-grained representations from earlier layers adaptively guide fine-grained feature extraction in subsequent layers via gated pathways. And ASR-MoE dynamically fuses multi-scale predictions by leveraging specialized global and local experts with temporal-aware weighting. Comprehensive experiments on thirteen real-world benchmarks demonstrate that DMSC consistently maintains state-of-the-art (SOTA) performance and superior computational efficiency for TSF tasks. Code is available at https://github.com/1327679995/DMSC.

Abstract

Explore similar work

DMSC: Dynamic Multi-Scale Coordination Framework for Time Series Forecasting

Generalizing Multi-Scale Time-Series Modeling with a Single Operator

SDMixer: Sparse Dual-Mixer for Time Series Forecasting