Network model optimization method and device, computer equipment and storage medium

Abstract

The invention relates to a network model optimization method and device, computer equipment and a storage medium. The method comprises the steps of obtaining a target population, the target population comprises a first number of individuals, and the individuals correspond to a set of model parameters of a to-be-trained network model; evolving the target population based on the evolution parameters, obtaining an evolved population, the evolved population comprises a second number of individuals, and the second number is larger than the first number; training the to-be-trained network model by using individuals in the evolved population to obtain a second number of trained individuals after environment change; obtaining a first number of environment-changed individuals from the second number of environment-changed individuals; and when the model optimization end condition is not met, updating the evolution parameters, taking the first number of individuals after environment change as a new target population, and returning to the step of evolving the target population based on the evolution parameters until the model optimization end condition is met. By adopting the method provided by the embodiment of the invention, the performance of the network model can be effectively improved.

Description

technical field [0001] The present application relates to the technical field of machine learning, in particular to a network model optimization method, device, computer equipment and storage medium. Background technique [0002] In industrial applications and scientific research, such as job shop scheduling, combination optimization, engineering design, power scheduling, investment management, image segmentation, network communication, data mining and other optimization fields, decision makers often encounter a class with multiple objectives and Time-varying optimization problems, such problems are often called dynamic multi-objective optimization problems. With the development of machine learning technology, the dynamic multi-objective optimization problem has gradually become one of the hot research topics. The main characteristic of the dynamic multi-objective optimization problem is the essential change. In the traditional optimization algorithm, most of them use the d...

AI Technical Summary

Problems solved by technology

[0003] However, the random selection of input parameters in the dynamic non-dominated sorting genetic algorithm will make the running trajectory of the optimization algorithm deviate. The gradient descent optimization loss function is a local search optimization algorithm, which can easily cause the network model to only converge to a local optimum. , thus leading to poor performance of the network model

Images