Exportar Publicação

Stefenon, S. F., Matos-Carvalho, J. P., Leithardt, V. R. Q. & Yow, K.-C. (2026). CNN-transformer-based model explained by SHAP and multi-head attention weights for time series forecasting. Scientific Reports, 16. https://doi.org/10.1038/s41598-026-63013-0

S. S. Frizzo et al.,  "CNN-transformer-based model explained by SHAP and multi-head attention weights for time series forecasting", in Scientific Reports, vol. 16, 2026

@article{frizzo2026_1794746092483,
	author = "Stefenon, S. F. and Matos-Carvalho, J. P. and Leithardt, V. R. Q. and Yow, K.-C.",
	title = "CNN-transformer-based model explained by SHAP and multi-head attention weights for time series forecasting",
	journal = "Scientific Reports",
	year = "2026",
	volume = "16",
	number = "",
	doi = "10.1038/s41598-026-63013-0",
	url = "https://www.nature.com/srep/"
}

TY  - JOUR
TI  - CNN-transformer-based model explained by SHAP and multi-head attention weights for time series forecasting
T2  - Scientific Reports
VL  - 16
AU  - Stefenon, S. F.
AU  - Matos-Carvalho, J. P.
AU  - Leithardt, V. R. Q.
AU  - Yow, K.-C.
PY  - 2026
SN  - 2045-2322
DO  - 10.1038/s41598-026-63013-0
UR  - https://www.nature.com/srep/
AB  - Convolutional Neural Networks (CNNs) and transformer architectures offer strengths for modeling temporal data: CNNs excel at capturing local patterns and translational invariances, while transformers effectively model long-range dependencies via self-attention. This paper proposes a hybrid architecture integrating convolutional feature extraction with a multi-head attention backbone to enhance multivariate time series forecasting. The CNN module first applies a hierarchy of one-dimensional convolutional layers to distill salient local patterns from raw input sequences, reducing noise and dimensionality. The resulting feature maps are then fed into the prediction model, which applies multi-head attention to capture both short- and long-term dependencies and to weigh relevant covariates adaptively. We evaluate the proposed CNN-transformer-based model on a hydroelectric natural flow time series dataset. Experimental results demonstrate that the proposed model outperforms well-established deep learning models, with a mean absolute percentage error of up to 2.2%. The explainability of the model is obtained by a proposed SHapley Additive exPlanations (SHAP) and Multi-Head Attention Weights (MHAW). Our novel architecture, named CNN-Transformer-SHAP-MHAW, is promising for applications requiring high-fidelity and multivariate time series forecasts.
ER  -