Randomized machine learning procedures

Yu. S. Popkov<sup>abcde

a</sup> Federal Research Center for Information Science and Control, Russian Academy of Sciences, Moscow, Russia
^b Trapeznikov Institute of Control Sciences, Russian Academy of Sciences, Moscow, Russia
^c Braude College of Haifa University, Karmiel, Israel
^d Yugra Research Institute of Information Technologies, Khanty-Mansiysk, Russia
^e Moscow Institute of Physics and Technology, Dolgoprudny, Russia

Abstract: A new concept of machine learning based on the computer simulation of entropy-optimal randomized models is proposed. The procedures of randomized machine learning (RML) with “hard” and “soft” randomization are considered; the former imply the exact reproduction of empirical balances while the latter their rough reproduction with an accepted approximation criterion. RML algorithms are formulated as functional entropy-linear programming problems. Applications of RML procedures to text classification and the randomized forecasting of migratory interaction of regional systems are presented.

Keywords: randomization, hard and soft randomization procedures, uncertainty, entropy, matrix norms, empirical balances, text classification, dynamic regression.


Received: 06.06.2018
Revised: 13.09.2018
Accepted: 08.11.2018

DOI: 10.1134/S0005231019090095

 English version: Automation and Remote Control, 2019, 80:9, 1653–1670

Bibliographic databases:

• 
• 
•