Hopfield Network, Application Of Genetic Algorithm In Data Mining

Recently,data mining arouses tremendous attention,for the main reason that a lot of data can be used and need to be transformed into useful information and knowledge.The information and knowledge obtained can be used in many fields,including business management,process control,market analysis,engineering design and science exploration and so on.In the past 30 years,steady and amazing progress in computer hardware has induced plentiful supply of powerful computer,data collection device and access media.Technology greatly promotes the development of database and information industry.More and more databases and information are used to transaction management information search and data analysis.To obtain important information from huge data to make decision,data mining emerges as the times require.Being an important branch of data mining,association rules mining can find interrelation of items.As a lot of data are collected and stored,people are more and more interested in mining association rules from their databases.There are several methods for association rules mining,apriori algorithm and some algorithms deriving from infrequent pattern growth method and so on.However,every method has its own flaw,people now try to improve existings and explore new approaches.Artificial neural network is a jumped-up science simulating structure and mechanism of human beings' brain.It is not comprehensive description of our brain but abstract simulation and predigestion,aiming to explore the way of process,access and search information to find a new approach to develop artificial intelligence and other sciences.Artificial neural network simulates principle of nerve cell,has ability of self-study,association, contrast, ratiocination and recapitulation.It supplies a new way to solve association rules mining.Recently,as one kind of artificial neural network, hopfield network is used for association rules mining and acquires remarkable results.However,existing methods have some flaws for coherent disadvantage of hopfield network.Genetic algorithm was first brought forward by John Holland in 1975. Genetic algorithm is a kind of adaptive search strategy.Because genetic algorithm is adaption, parallelism,and good at dealing with huge data,it is widely used in many fields.Therefore,genetic algorithm is used to train artificial neural network, to improve topology and weight of network.In practice,application of genetic algorithm to evolve BP network is more and more mature and gets good results.In contrast,though some achievements appear,combination of genetic and hopfield network need to be prospered.This paper dedicates to combine hopfield network,genetic algorithm and association rules mining together, introduce a new idea that use genetic algorithm locally to evolve hopfield network and then use the network to mine association rules.