Search International and National Patent Collections

1. (WO2018161468) GLOBAL OPTIMIZATION, SEARCHING AND MACHINE LEARNING METHOD BASED ON LAMARCK ACQUIRED GENETIC PRINCIPLE

Pub. No.:    WO/2018/161468    International Application No.:    PCT/CN2017/089285
Publication Date: Fri Sep 14 01:59:59 CEST 2018 International Filing Date: Thu Jun 22 01:59:59 CEST 2017
IPC: G06N 3/12
Applicants: DONGGUAN UNIVERSITY OF TECHNOLOGY
东莞理工学院
LI, Yun
李耘
LI, Lin
李琳
Inventors: LI, Yun
李耘
LI, Lin
李琳
Title: GLOBAL OPTIMIZATION, SEARCHING AND MACHINE LEARNING METHOD BASED ON LAMARCK ACQUIRED GENETIC PRINCIPLE
Abstract:
A global optimization, searching and machine learning method based on the Lamarck acquired genetic principle, comprising: step 1: constructing an objective function f(x) according to a problem object; step 2: encoding the problem object into a chromosome of a genetic algorithm, automatically calculating or inputting operation parameters, and initializing same; step 3: performing iteration and optimization on a current (k-th generation) population Gk={P1k, P2k,...Psk} by means of a Lamarck "acquired genetic operator" and "a use and disuse operator" of the invention according to an evaluation of the objective function f(x); and step 4: inputting an optimal solution set of the problem object. In the method, combining the natural laws of "acquired genetics" and "use and disuse" of Lamarck's theory of evolution with modern "epigenetics" and the natural law of "survival of the fittest" of Darwin's theory of evolution simplifies the structure of genetic algorithms, overcomes multiple technical defects of existing algorithms, and improves the efficiency, global optimality, and sustainability of later evolution thereof, so as to better solve more problems regarding global optimization, searching and machine learning.