Reviving Error Correction in Modern Deep Time-Series Forecasting

Minh Hoang Nguyen, Van Dai Do, Huu Hiep Nguyen, Dung Nguyen, Kien Do, Hung Le
Proceedings of the 43rd International Conference on Machine Learning, PMLR 306:92824-92850, 2026.

Abstract

Modern deep-learning models have achieved remarkable success in time-series forecasting. Yet, their performance degrades in long-term prediction due to error accumulation in autoregressive inference, where predictions are recursively used as inputs. While classical error correction mechanisms (ECMs) have long been used in statistical methods, their applicability to deep learning models remains limited or ineffective. In this work, we revisit the error accumulation problem in deep time-series forecasting and investigate the role and necessity of ECMs in this new context. We propose a simple, architecture-agnostic error correction model that can be integrated with any existing forecaster without requiring retraining. By explicitly decomposing predictions into trend and seasonal components and training the corrector to adjust each separately, we introduce the Universal Error Corrector with Seasonal–Trend Decomposition (UEC-STD), which significantly improves correction accuracy and robustness across 4 backbones and 10 datasets. Our findings provide a practical tool for enhancing forecasts while offering new insights into mitigating autoregressive errors in deep time-series models.

Cite this Paper


BibTeX
@InProceedings{pmlr-v306-nguyen26u, title = {Reviving Error Correction in Modern Deep Time-Series Forecasting}, author = {Nguyen, Minh Hoang and Do, Van Dai and Nguyen, Huu Hiep and Nguyen, Dung and Do, Kien and Le, Hung}, booktitle = {Proceedings of the 43rd International Conference on Machine Learning}, pages = {92824--92850}, year = {2026}, editor = {Zhang, Tong and Dudik, Miroslav and Jaggi, Martin and Agarwal, Alekh and Li, Sharon and Schuurmans, Dale and Zhu, Jerry and Berkenkamp, Felix and Dong, Hanze and Bietti, Alberto}, volume = {306}, series = {Proceedings of Machine Learning Research}, month = {06--11 Jul}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v306/main/assets/nguyen26u/nguyen26u.pdf}, url = {https://proceedings.mlr.press/v306/nguyen26u.html}, abstract = {Modern deep-learning models have achieved remarkable success in time-series forecasting. Yet, their performance degrades in long-term prediction due to error accumulation in autoregressive inference, where predictions are recursively used as inputs. While classical error correction mechanisms (ECMs) have long been used in statistical methods, their applicability to deep learning models remains limited or ineffective. In this work, we revisit the error accumulation problem in deep time-series forecasting and investigate the role and necessity of ECMs in this new context. We propose a simple, architecture-agnostic error correction model that can be integrated with any existing forecaster without requiring retraining. By explicitly decomposing predictions into trend and seasonal components and training the corrector to adjust each separately, we introduce the Universal Error Corrector with Seasonal–Trend Decomposition (UEC-STD), which significantly improves correction accuracy and robustness across 4 backbones and 10 datasets. Our findings provide a practical tool for enhancing forecasts while offering new insights into mitigating autoregressive errors in deep time-series models.} }
Endnote
%0 Conference Paper %T Reviving Error Correction in Modern Deep Time-Series Forecasting %A Minh Hoang Nguyen %A Van Dai Do %A Huu Hiep Nguyen %A Dung Nguyen %A Kien Do %A Hung Le %B Proceedings of the 43rd International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2026 %E Tong Zhang %E Miroslav Dudik %E Martin Jaggi %E Alekh Agarwal %E Sharon Li %E Dale Schuurmans %E Jerry Zhu %E Felix Berkenkamp %E Hanze Dong %E Alberto Bietti %F pmlr-v306-nguyen26u %I PMLR %P 92824--92850 %U https://proceedings.mlr.press/v306/nguyen26u.html %V 306 %X Modern deep-learning models have achieved remarkable success in time-series forecasting. Yet, their performance degrades in long-term prediction due to error accumulation in autoregressive inference, where predictions are recursively used as inputs. While classical error correction mechanisms (ECMs) have long been used in statistical methods, their applicability to deep learning models remains limited or ineffective. In this work, we revisit the error accumulation problem in deep time-series forecasting and investigate the role and necessity of ECMs in this new context. We propose a simple, architecture-agnostic error correction model that can be integrated with any existing forecaster without requiring retraining. By explicitly decomposing predictions into trend and seasonal components and training the corrector to adjust each separately, we introduce the Universal Error Corrector with Seasonal–Trend Decomposition (UEC-STD), which significantly improves correction accuracy and robustness across 4 backbones and 10 datasets. Our findings provide a practical tool for enhancing forecasts while offering new insights into mitigating autoregressive errors in deep time-series models.
APA
Nguyen, M.H., Do, V.D., Nguyen, H.H., Nguyen, D., Do, K. & Le, H.. (2026). Reviving Error Correction in Modern Deep Time-Series Forecasting. Proceedings of the 43rd International Conference on Machine Learning, in Proceedings of Machine Learning Research 306:92824-92850 Available from https://proceedings.mlr.press/v306/nguyen26u.html.

Related Material