Lattice Theory for Machine Learning
Dmitriy V. Vinogradov · Scientific and Technical Information Processing · 2022
Abstract This article presents a theoretical foundation for constructing a machine learning system based on the binary similarity operation. The key technique used is formal concept analysis, a modern branch of lattice theory. Algorithms for encoding objects with both discrete and continuous attributes are introduced, the Markov chain Monte Carlo method is described, and the key steps of machine learning are discussed. Finally, results that demonstrate a sufficient number of generated hypotheses are presented. The results of experimental testing of the proposed approach on several datasets from the UCI Machine Learning repository are also discussed.