Google Colaboratory as a Tool in Remote Teaching of Machine Learning I: Challenges, Solutions, and Results

• Martony Demes da Silva UFPI
• Nara Leyla dos Santos Silva UFPI

DOI: https://doi.org/10.5753/ctrle.2025.10763

Abstract

This article presents an experience report on the use of Google Colaboratory (Colab) in remote teaching of the Machine Learning I course at the University UF. The transition to remote learning required the adaptation of methodologies and tools, and Google Colaboratory was chosen as the platform to enable the practical activities of the course, allowing students to work with machine learning algorithms using interactive notebooks and cloud-based computational resources. The article discusses the main challenges faced, such as the limitation of free GPUs, students' connectivity issues, and the need for model optimization. It also presents the solutions adopted, including the use of pre-trained models, flexible task structures, and the creation of collaborative forums. Preliminary results indicate that the platform was effective in facilitating hands-on learning in machine learning, offering an interactive and accessible experience, albeit with challenges related to infrastructure and connectivity. The article concludes with suggestions for future improvements, including the use of other cloud platforms and the implementation of more interactive pedagogical strategies.

References

Lima, A. S.; Pacheco, V. T. (2022). Potencialidades do Google Colab no ensino de programação e ciência de dados. Revista Brasileira de Informática na Educação (RBIE), 30(2), 112–130. DOI: 10.5753/rbie.2022.30.2.112

Oliveira, T. G.; Lima, E. M.; Costa, P. H. (2021). Barreiras e estratégias no uso do Google Colab no ensino de computação. In Anais do Congresso Brasileiro de Informática na Educação (CBIE) (Vol. 1, No. 1, pp. 312–321). Porto Alegre: Sociedade Brasileira de Computação. DOI: 10.5753/cbie.2021.312

Santos, F. P. dos; Lima, A. C.; Moura, D. G. (2020). Tecnologias educacionais emergentes no ensino de programação: desafios e possibilidades. Revista Tecnologias na Educação, 12(30), 45–59. DOI: 10.9771/tecined.v12i30.37786

Silva, M. T.; Gomes, R. A. (2021). Uso de modelos pré-treinados no ensino de redes neurais com recursos computacionais limitados. In Anais do Workshop sobre Tecnologias na Educação (Ctrl+E) (Vol. 1, pp. 75–84). Porto Alegre: Sociedade Brasileira de Computação. DOI: 10.5753/ctrle.2021.75