Aprendizagem Profunda e Inteligência Artificial Verde: Caminhos para um Futuro mais Sustentável Resumo Estendido – CTDG-SI 2026
Resumo
In the last decade, there have been significant advances in the results achieved by Deep Learning models and their widespread adoption in academia and industry. Although these models have the potential to assist in the management of natural resources and environmental issues, they typically require a great deal of computing power, resulting in higher energy costs and also large carbon footprint numbers. This work seeks to highlight and discuss the energy costs involved in using neural network models, experimentally comparing some architectures in terms of performance, energy efficiency, and computational cost. The results obtained reinforce that it is possible to build models that reconcile sustainability and performance, providing subsidies for more conscious technical choices.
Referências
Lecun, Y., Bottou, L., Bengio, Y., and Haffner, P. (1998). Gradient-based learning applied to document recognition. Proceedings of the IEEE, 86(11):2278–2324.
Lenherr, N., Pawlitzek, R., and Michel, B. (2021). New universal sustainability metrics to assess edge intelligence. Sustainable Computing: Informatics and Systems, 31:100580.
Schwartz, R., Dodge, J., Smith, N. A., and Etzioni, O. (2020). Green ai. Communications of the ACM, 63(12):54–63.
Wolff Anthony, L. F., Kanding, B., and Selvan, R. (2020). Carbontracker: Tracking and predicting the carbon footprint of training deep learning models. arXiv.org.
