An optimization model method and application based on generative confrontation network

Abstract

The present invention discloses an optimization model method and application based on a generative confrontation network, called GAN-O, including steps: expressing the application (such as logistics distribution optimization) as a function optimization problem; according to the test function and test dimension of the function optimization problem , establish a function optimization model based on the generative confrontation network, including building a generator and a discriminator based on the generative confrontation network; train the function optimization model; use the trained function optimization model to perform iterative calculations to obtain the optimal solution; optimization solution. The invention can obtain a better local optimal solution in a shorter time, makes the training of the deep neural network stable, and has better local search ability. The present invention can be applied to problems in many application scenarios that can be transformed into function optimization problems in reality, such as logistics distribution problems, etc., has a wide range of applications, can solve a large number of practical problems, and has great promotion and application value.

Description

technical field [0001] The invention relates to the technical field of computational model optimization, in particular to a novel optimization model method and application based on a generative confrontation network. Background technique [0002] Function optimization problems have always been one of the most important problems in the fields of mathematics and computer science. In reality, many application scenarios can be transformed into function optimization problems, such as logistics distribution problems, deep network optimization problems, etc. The application field of function optimization problem is very broad, and it can solve a large number of practical problems. [0003] For function optimization problems, the existing algorithms are mainly gradient-based algorithms. The disadvantage of this type of algorithm is that it is very easy to fall into local extremum. For some problems, such as neural network optimization, the local extremum usually has a good enough e...

AI Technical Summary

Problems solved by technology

[0007] In order to overcome the shortcomings of the above-mentioned prior art, the present invention provides a new optimization model method and application based on generative confrontation network, which can be used to search for the global optimal solution of continuous functions, and solve the lack of diversity in the local search of existing function optimization algorithms The problem

Images