Towards Green PHM: Adaptive Early Stopping for Sustainable Neural Architecture Search in Industrial Applications
##plugins.themes.bootstrap3.article.main##
##plugins.themes.bootstrap3.article.sidebar##
Abstract
Neural Architecture Search has revolutionized Prognostics and Health Management, yet adoption is often hindered by the massive carbon footprint generated during the evaluation of candidate architectures. To address this sustainability challenge, this work introduces a Green AI framework that significantly reduces the energy consumption of such searches through an intelligent and adaptive early stopping mechanism. The approach utilizes Prototypical Networks for regression to extrapolate learning curves from partial data, predicting final model performance early in the training process.
A distinct sustainability advantage of using Prototypical Networks lies in the inherent data efficiency and superior generalization capabilities of the architecture. By leveraging metric learning, the framework avoids the energy intensive process of training task specific predictors from scratch. Instead, it enables few shot transfer across diverse domains, minimizing the total computational overhead of the search process. Furthermore, a key contribution of this framework is the dynamic adaptation of the decision logic as the optimization process evolves. By utilizing a decision tree classifier that adjusts thresholds based on the progression of the search, the system becomes increasingly selective and prioritizes computational resources for the most promising candidates.
\textcolor{blue}{The proposed framework was validated across sixty one thousand learning curves from fifty diverse datasets. Experimental results demonstrate a drastic reduction in total computational hours, achieving 57\% of decrease in training time while maintaining high diagnostic fidelity. The system consistently identified top tier model configurations, reaching a mean selection rank of 0.9 across tested industrial scenarios. These results prove that high performance industrial intelligence can be achieved without the prohibitive environmental costs typically associated with large scale architecture optimization.
How to Cite
##plugins.themes.bootstrap3.article.details##
learning curves, predictive maintenance, early stopping, green AI, Bayesian Optimization
Baker, B., Gupta, O., Raskar, R., & Naik, N. (2018). Accelerating neural architecture search using performance prediction.
Borrego-Díaz, J., & Galán-Páez, J. (2022). Explainable artificial intelligence in data science. Minds and Machines, 32(3), 485–531. doi: 10.1007/s11023-022-09603-z
Domhan, T., Springenberg, J. T., & Hutter, F. (2015). Speeding up automatic hyperparameter optimization of deep neural networks by extrapolation of learning curves. In Proceedings of the 24th International Conference on Artificial Intelligence (pp. 3460–3468). AAAI Press.
Egele, R., Mohr, F., Viering, T., & Balaprakash, P. (2024). The unreasonable effectiveness of early discarding after one epoch in neural network hyperparameter optimization. Neurocomputing, 127964.
Falkner, S., Klein, A., & Hutter, F. (2018). BOHB: Robust and efficient hyperparameter optimization at scale. In International Conference on Machine Learning (pp. 1437–1446).
Kandasamy, K., Neiswanger, W., Schneider, J., Poczos, B., & Xing, E. P. (2018). Neural architecture search with Bayesian optimisation and optimal transport. Advances in Neural Information Processing Systems, 31.
Klein, A., Falkner, S., Springenberg, J., & Hutter, F. (2017b). Learning curve prediction with Bayesian neural networks. In Proceedings of the International Conference on Learning Representations.
Klein, A., Falkner, S., Springenberg, J. T., & Hutter, F. (2017a). Learning curve prediction with Bayesian neural networks. In International Conference on Learning Representations.
Li, L., Jamieson, K. G., DeSalvo, G., Rostamizadeh, A., & Talwalkar, A. (2017). Hyperband: Bandit-based configuration evaluation for hyperparameter optimization. In ICLR Poster (p. 53).
Mettes, P., Van der Pol, E., & Snoek, C. (2019). Hyperspherical prototype networks. Advances in Neural Information Processing Systems, 32.
Real, E., Aggarwal, A., Huang, Y., & Le, Q. V. (2019). Regularized evolution for image classifier architecture search. In Proceedings of the AAAI Conference on Artificial Intelligence (Vol. 33, pp. 4780–4789).
Snell, J., Swersky, K., & Zemel, R. (2017). Prototypical networks for few-shot learning. Advances in Neural Information Processing Systems, 30.
Snoek, J., Larochelle, H., & Adams, R. P. (2012). Practical Bayesian optimization of machine learning algorithms. Advances in Neural Information Processing Systems, 25.
Solis-Martin, D., Galan-Paez, J., & Borrego-Diaz, J. (2024). Bayesian model selection pruning in predictive maintenance. In International Conference on Hybrid Artificial Intelligence Systems (pp. 263–274).
Solís-Martín, D., Galán-Páez, J., & Borrego-Díaz, J. (2025). A model for learning-curve estimation in efficient neural architecture search and its application in predictive health maintenance. Mathematics, 13(4), 555.
Solís-Martín, D., Galán-Páez, J., & Borrego-Díaz, J. (2025). PHMD: An easy data access tool for prognosis and health management datasets. SoftwareX, 29, 102039. doi: https://doi.org/10.1016/j.softx.2025.102039
Zhang, W., Peng, G., Li, C., Chen, Y., & Zhang, Z. (2017). A new deep learning model for fault diagnosis with good anti-noise and domain adaptation ability on raw vibration signals. Sensors, 17(2), 425.
Zoph, B., Vasudevan, V., Shlens, J., & Le, Q. V. (2018). Learning transferable architectures for scalable image recognition. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (pp. 8697–8710).
This work is licensed under a Creative Commons Attribution 3.0 Unported License.
The Prognostic and Health Management Society advocates open-access to scientific data and uses a Creative Commons license for publishing and distributing any papers. A Creative Commons license does not relinquish the author’s copyright; rather it allows them to share some of their rights with any member of the public under certain conditions whilst enjoying full legal protection. By submitting an article to the International Conference of the Prognostics and Health Management Society, the authors agree to be bound by the associated terms and conditions including the following:
As the author, you retain the copyright to your Work. By submitting your Work, you are granting anybody the right to copy, distribute and transmit your Work and to adapt your Work with proper attribution under the terms of the Creative Commons Attribution 3.0 United States license. You assign rights to the Prognostics and Health Management Society to publish and disseminate your Work through electronic and print media if it is accepted for publication. A license note citing the Creative Commons Attribution 3.0 United States License as shown below needs to be placed in the footnote on the first page of the article.
First Author et al. This is an open-access article distributed under the terms of the Creative Commons Attribution 3.0 United States License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.