Neural Emulators in Identification of Order Parameters in Human Ecology

Abstract

Binary classification offers the procedure of identification of more significant x i diagnostic features (order parameters) for the systems being in different conditions and divided by a neural emulator. Ranking of x i is realized by many repetitions (p > 100) of problem solving of binary classification by a neural emulator at initial set of feature weights w j io where j is an iteration number (i = 1, 2 ... p) from the interval (0, 1) with an uniform distribution. Accuracy of identification increases by increased p (p > 1 000, p > 10 4), that is demonstrated on parameters of cardiovascular system of groups of people working at night shifts.