Research On Hierarchical Co-evolution Model And Algorithm For Fashion Color Prediction

Research shows that ecosystem is a biology system with a level of evolution pattern of natural systems.The population is made by individuals,Species are made by populations,and the biological system is made by populations.There is not only a competition for resources but also a collaboration relationship for overcoming bad environment.Meanwhile,the phenomenons from individuals to populations and from populations to communities reflect the low level to high level of swarm intelligence individual behavior The evolution,symbiotic evolution,complex adaptive system theory as its important theoretical support are regards as the important theoretical supports in biological heuristic computation.In recent years,the scholars searching biological heuristic computation focus on the community information and complex symbiotic relations from simple biological behavior description to the communication patterns between species.This paper focuses on the symbiotic evolution phenomenon and hierarchical information dissemination phenomenon in biological systems.After studying the individual information communication and adaptive adjustment rules,this paper s a co-evolution mechanism with a level of unified framework model.With the help of the proposed framework,the classical artificial bee colony algorithm is improved into hierarchical multi-hive algorith with a well performance,which retains the good characteristics such as high convergence rate and overcomes the fast rate of loosing population diversity.Then the improved algorithm is applied in calculating the weight values of artificial netural network.This network is applied in predicting the hue of the fashion color in next years.This method can check the validity and effectiveness in solving practical problems.In this paper,the main research work is as follows:In this paper,a model based on hierarchical synergy evolution model with multiple species evolution is proposed.This model makes reference to the theory of complex adaptive system and collaborative evolutionary.This model ranges from the individual to the hierarchy of species to the community.After retaining the the mode of information communication between levels and competition-collaborative symbiosis model tree within the same level,it can be realized as the intelligence between the individual layer,population layer and community layer.(1)In this paper,a multiple colony collaborative optimization algorithm based on community level evolution is proposed according to individual-populations-species collaboration evolution mode with multiple hives.By information communication mode and peer competition-collaborative symbiotic within hierarchy model,it is interlligence among the individual layer,population layer and community layer.(2)Based on the multi-populaiton collaboration framework with three layers-individuals,populations and speices,this paper propose a hierarchical multi-hive artificial bee coloy algorithm(HABC).The divide-and-conquer stratergy is applied in reducing the difficult in soving high-dimension problems.The characteristic vector can be divided into several short vectors stochastically.Crossover operation and elite strategy can enhance information exchange between populations,in order to keep the population diversity.Crossover operation and elite strategy to enhance information exchange between populations,in order to keep the population diversity.15 continuous functions and 5 discrete test functions are choosen.Test results show that HABC at the beginning of the optimization process can maximum limit to keep population diversity and have a faster convergence rate in the late optimization.Compared with some classical heuristic evolutionary algorithm,this algorithm not only has a high optimization precision,and also has high ruban yogarajah,especially for high dimensional complex problems,its performance is more outstanding.(3)To overcome the drawbacks such as easily trapping local minimum and high computational complexity,precision dependant on initially selection of parameters of artificial netural network,the classical gradient descent or value optimizations is unable.The hierarchical multi-hive collaboration algorithm is applied in calculating the weight values of network.After testing on contiouns benchmark functions and discrete functions,the results show that the HABC algorithm can have a higher convergence rate and a higher learning effectness than classical BP netural network learning method and classical artificial bee colony algorithm.(4)In predicting the clothes fashion color,the quantitative and classification of hue would affect the prediction precision.According to the color science theory,the hue wheel of PANTONE is divided.Interval classification of boundary value is proposed.It is also an important basis for digitizing the the subjective color hue.Experiment from 2007 to 2018 released by the international commission on color women international spring and summer clothing fashion color finalized as the research data,with the help of evolution based on community level HABC collaborative algorithm for neural network,consumers purchase intention of spring and summer dress popular colour is simulated.