Randomization and entropy in machine learning and data processing

Combining the concept of randomization with entropic criteria allows solutions to be obtained in the conditions of maximum uncertainty, which is very effective in machine learning and data processing. The application of this approach in data-based entropy-randomized evaluation of functions, randomized hard and soft machine learning, object clustering, and data matrix dimension reduction is demonstrated. Some applications of classification problems, forecasting the electric load of a power system, and randomized clustering of biological objects are considered.