Reflection of Measurable Features in an Artificial Intelligence System
摘要
Abstract
Measured characteristics are characterized by uncertainty due to possible deviations of the initial values from the true ones. To overcome this effect, it is proposed to use the probability distribution of values. This will increase the reliability of the hypotheses generated by intelligent systems. Formalization of the input sequence of numerical features can be implemented in different ways. An extended representation of a slot in a frame for features with quantitative values is considered.