Self-Organizing Maps and Autoencoders
Abstract
This paper presents an idea that combines SOM (Self-Organizing Map) and convolutional autoencoders to create a hierarchical and interpretable model. The proposed method replaces each neuron in the SOM grid with an autoencoder. The SOM organizes the grid topologically, creating neighborhood relationships among the autoencoders. Each autoencoder serves as a feature extractor, as its encodings store the main features of an input in a latent representation. In a separate training phase, with the SOM training already completed, the latent layers generated by the autoencoders are used in Multilayer Perceptron (MLP) networks for image classification tasks.
References
Bank, D., Koenigstein, N., and Giryes, R. (2023). Autoencoders. In Rokach, L., Maimon, O., and Shmueli, E., editors, Machine Learning for Data Science Handbook. Springer.
Byerly, A., Kalganova, T., and Dear, I. (2021). No routing needed between capsules. Neurocomputing, 463:545–553.
Géron, A. (2021). Aprendizado de representação e aprendizado gerativo com autoencoders e gans. In Mãos à Obra: Aprendizado de Máquina com Scikit-Learn, Keras & TensorFlow, chapter 17, pages 437–446. Alta Books, 2nd edition.
Huijben, I., Mettes, P., and Snoek, C. G. M. (2023). Som-cpc: Unsupervised contrastive learning with self-organizing maps. IEEE Transactions on Pattern Analysis and Machine Intelligence. Early Access.
Jogin, M. et al. (2018). Feature extraction using convolution neural networks (cnn) and deep learning. In 2018 3rd IEEE International Conference on Recent Trends in Electronics, Information & Communication Technology (RTEICT), pages 2319–2323. IEEE.
Khacef, A., Hébrail, G., and Gallinari, P. (2020). Improving self-organizing maps with unsupervised feature extraction. In 2020 International Joint Conference on Neural Networks (IJCNN), pages 1–8. IEEE.
Kohonen, T. (1982). Self-organized formation of topologically correct feature maps. Biological Cybernetics, 43(1):59–69.
Kriesel, D. (2007). Self-organizing feature maps. In A Brief Introduction to Neural Networks, chapter 10, pages 147–164. Disponível em: [link].
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.
Byerly, A., Kalganova, T., and Dear, I. (2021). No routing needed between capsules. Neurocomputing, 463:545–553.
Géron, A. (2021). Aprendizado de representação e aprendizado gerativo com autoencoders e gans. In Mãos à Obra: Aprendizado de Máquina com Scikit-Learn, Keras & TensorFlow, chapter 17, pages 437–446. Alta Books, 2nd edition.
Huijben, I., Mettes, P., and Snoek, C. G. M. (2023). Som-cpc: Unsupervised contrastive learning with self-organizing maps. IEEE Transactions on Pattern Analysis and Machine Intelligence. Early Access.
Jogin, M. et al. (2018). Feature extraction using convolution neural networks (cnn) and deep learning. In 2018 3rd IEEE International Conference on Recent Trends in Electronics, Information & Communication Technology (RTEICT), pages 2319–2323. IEEE.
Khacef, A., Hébrail, G., and Gallinari, P. (2020). Improving self-organizing maps with unsupervised feature extraction. In 2020 International Joint Conference on Neural Networks (IJCNN), pages 1–8. IEEE.
Kohonen, T. (1982). Self-organized formation of topologically correct feature maps. Biological Cybernetics, 43(1):59–69.
Kriesel, D. (2007). Self-organizing feature maps. In A Brief Introduction to Neural Networks, chapter 10, pages 147–164. Disponível em: [link].
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.
Published
2025-08-12
How to Cite
RODRIGUES, Elias C.; MATOS, Leonardo N..
Self-Organizing Maps and Autoencoders. In: REGIONAL SCHOOL ON COMPUTING OF BAHIA, ALAGOAS, AND SERGIPE (ERBASE), 25. , 2025, Lagarto/SE.
Anais [...].
Porto Alegre: Sociedade Brasileira de Computação,
2025
.
p. 281-287.
DOI: https://doi.org/10.5753/erbase.2025.13757.
