A deep machine learning algorithm for construction of the Kolmogorov–Arnold representation

The Kolmogorov–Arnold representation is a proven adequate replacement of continuous multivariate function by hierarchical structure multiple functions one variable. existence such inspired many researchers to search for practical way its construction, since model answers the needs machine learning. This article shows that not only composition but also particular case tree discrete Urysohn operators. introduces new, quick and computationally stable algorithm constructing trees. Besides functions, suggested covers cases with quantised inputs combination inputs. contains results testing on publicly available datasets, used other benchmarking.