Multilevel recursive model of properties existence constraints in machine learning
D E Samoylov, Valentina Semenova, Sergey V. Smirnov · Journal of Physics Conference Series · 2018
We considered an a priori knowledge that establishes how the facts of detection some properties in empirically observed objects affect the presence of others. Taking into account such constraints of the existence of properties in the objects of the domain of interest is necessary for the correct formation of the contexts of the tasks of machine learning in the conditions of incompleteness and inconsistency of the initial data. The paper presents a new multilevel recursive model of such constraints. The model is built on the basis of the point of view on the logical and conceptual origin of restrictions using elementary binary relations on a set of measured properties - incompatibility and conditionality.