Population generation method and selection method and device of neural network architecture

Abstract

The invention relates to a population generation method, a selection method and selection device of a neural network architecture. The method comprises the steps of obtaining a t-th generation population Rt; carrying out non-dominated sorting on individuals contained in the population Rt to obtain k individual groups with different ranks; sequentially selecting i individual groups from the k individual groups; calculating the weighted congestion distance of the jth individual under the plurality of targets according to the congestion distance of the jth individual under each target and the weight corresponding to each target; selecting m individuals from the ith individual group according to the weighted congestion distance, and forming a (t+1)th generation population R<t+1> together withall individuals in the (i-1)th individual group; and setting t to be equal to t+1, starting to execute the step of carrying out non-dominated sorting on individuals contained in the population Rt again, and when a Tth generation population RT is generated, determining that the population RT is a target population. The population meeting the preference of the decision maker is generated, and the flexibility of determining the individual congestion degree is improved.

Description

technical field [0001] The embodiments of the present disclosure relate to the field of computer technology, and in particular to a population generation method, a method and a device for selecting a neural network architecture. Background technique [0002] NSGA-II (Non-dominated Sorting Genetic Algorithm-II, non-dominated sorting genetic algorithm with elite strategy) is an improved algorithm for NSGA (Non-dominated Sorting Genetic Algorithm, non-dominated sorting genetic algorithm). [0003] In NSGA-II, each individual in the population has two attributes: non-dominated sort attribute and crowding degree attribute. Small individuals will be reserved preferentially; when two individuals have the same non-dominated sort attributes, that is, when the non-dominated sorts are the same, the crowding property is better, that is, individuals with less crowded surroundings will be preferentially retained. [0004] However, for individuals with the same non-dominated sorting attri...

AI Technical Summary

Problems solved by technology

[0004] However, for individuals with the same non-dominated sorting attributes, the crowding degree attribute is determined according to the crowding degree of the individual on different optimization objectives, and the determination method is relatively simple

Images