The Study of Retraining in Algebraic Machine Learning
L. A. Iakimova · Pattern Recognition and Image Analysis · 2023
Abstract The article describes the results of an experimental study of the behavior of solvers based on the binary similarity operation. As phenomena of interest to us, the classical problem of retraining (especially its stability with an increase in the minimum number of generating examples) and the resulting concept of phantom similarities in data mining systems are investigated.