In this paper, we propose Deep Extremely Randomized Trees (DET), a deep extension of the extremely randomized trees (Extra-Trees) approach. Our approach unifies classification trees with the layered learning method inspired from deep neural networks. The DET is a deep structure where each layer is a set of Extra-Trees. We look at experimental results on machine learning gold standard datasets and find on-par or superior results when compared to state-of-the-art deep models, with much less parameters to tune. It performs faster than deep neural classifiers, and in most cases even faster than gcForest, without losing accuracy.
CITATION STYLE
Berrouachedi, A., Jaziri, R., & Bernard, G. (2019). Deep extremely randomized trees. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) (Vol. 11953 LNCS, pp. 717–729). Springer. https://doi.org/10.1007/978-3-030-36708-4_59
Mendeley helps you to discover research relevant for your work.