Genetic Algorithm To Optimize Neural Network Application In Recommendation System

With the rapid development of information technology,the Internet has become an indispensable tool in people's daily life.However,the commercial Internet site structure is getting more and more complicated,which makes a problem that users frequently lost themselves in a large number of commodity information.E-commerce recommendation system arises at the historic moment,it is one of the most effective tool that is to solve the'information overload' problem.It has become a reality that people could use artificial intelligence system to predict and analyze user's preference,and recommend to users of information.This study is to apply genetic algorithm optimize neural network to the electronic commerce recommendation platform.The study proceed data mining based on the usersr search information,browse information,click information,order information and evaluate information,and analyze their behaviors by their interactive mode,and predict their preference to recommend to them of goods and services that they are interested in.In the era of electricity,companies could provide better services to customers with personalized recommendation system,which can improve the user's satisfaction and loyalty,it can provide more entries of the goods,and let the enterprise get core advantages in today's fierce market competition,even improve enterprises' sales levels.This paper particularly describes the basic theory of genetic algorithm and artificial neural network,analyze and improve the important artificial intelligence algorithm,and it tries to fuse and analyze the genetic algorithm neural network structure and parameters optimization.By modifying the initial values and parameters,the new genetic algorithm has improved the stability and efficiency of the neural network,and it will get more effective application in the field of electronic commerce.This article is mainly about:Firstly,introduction to the related concepts and working principles of genetic algorithm,enumeration of the advantages and disadvantages of traditional genetic algorithm,and introduction to the problems that will be solved in the design of the application of genetic algorithm,such as fitness distribution,genetic operation,etc.Secondly,the study of the neural network basic algorithm and characteristics,and the advantages and disadvantages of the BP neural network,and the analysis of some the local minimum problems of BP neural network that may appear,and the correction proposals were putted forward,and the improvement of introducing the Levenberg—Marquardt.Thirdly,the application of genetic algorithm which was used to optimize the BP neural network,and the optimal neural network weights and threshold value is obtained,thus the LMBP neural network combination algorithm based on genetic algorithm was putted forward,which fully shows a global search ability of genetic algorithm,and it makes the algorithm forecast more accurate with BP neural network prediction ability.At the last,the combination of the improved algorithm GA-LMBP is applied to the field of electronic commerce recommendation,according to the customer's personal information and behavior data,it predict the commodities that customers may be interested in,and recommend it to the customers.This article choose films as an example,the simulation experiment system,and the better recommendation results have been achieved.