Cálculo paralelo de índice de validação de agrupamento

  • Vinicius F. Barbosa UFG
  • Wellington S. Martins UFG

Abstract


In meta-learning, meta-features are measures derived from the dataset that provide additional information about its properties. Extracting, or calculating, these meta-features can be costly, especially in clustering. This paper seeks to use parallelism to more quickly and efficiently compute a meta-feature, or index, of cluster validation. Our experiments show gains of up to 22 times compared to the sequential version.

References

Lemke, C., Budka, M., and Gabrys, B. (2015). Metalearning: a survey of trends and technologies. Artificial intelligence review, 44:117–130.

Luna-Romera, J. M., del Mar Martinez-Ballesteros, M., Garcia-Gutierrez, J., and Riquelme-Santos, J. C. (2016). An approach to silhouette and dunn clustering indices applied to big data in spark. In Advances in Artificial Intelligence: 17th Conference of the Spanish Association for Artificial Intelligence, CAEPIA 2016, Salamanca, Spain, September 14-16, 2016. Proceedings 17, pages 160–169. Springer.

Rivolli, A., Garcia, L. P., Soares, C., Vanschoren, J., and de Carvalho, A. C. (2018). Characterizing classification datasets: a study of meta-features for meta-learning. arXiv preprint arXiv:1808.10406.

Silva L., Franco R., C. A. M. W. (2023). Gpu acceleration of clustering meta-feature extraction using rapids. XXII Workshop em Desempenho de Sistemas Computacionais e de Comunicação, (wperformance 2023) edition.

Zerabi, S., Meshoul, S., and Boucherkha, S. C. (2020). Models for internal clustering validation indexes based on hadoop-mapreduce. International Journal of Distributed Systems and Technologies (IJDST), 11(3):42–67.
Published
2023-08-24
BARBOSA, Vinicius F.; MARTINS, Wellington S.. Cálculo paralelo de índice de validação de agrupamento. In: REGIONAL HIGH PERFORMANCE SCHOOL OF THE MIDWEST (ERAD-CO), 6. , 2023, Evento Online. Anais [...]. Porto Alegre: Sociedade Brasileira de Computação, 2023 . p. 20-22. DOI: https://doi.org/10.5753/eradco.2023.233552.