Based On The Context Of Genetic K-means Clustering Algorithm Model Of Quantitative Research

To statistics the correlation of high-order the source effectively, the Context model has been widely used.But it faces a serious problem-"model cost". Not providing enough data in the learning process, making the probability of the current coding symbol be unreliable. To solve this problem, Context quantization is introducted in the limited conditions of lossless image coding. Context quantization makes the current coding symbol probability statistics to be easier and more accurate, and solve the problem of the "model price", get the best coding.Many experiments have proved that the distortion criterion be selected in the right, the Context quantization is similar to vector quantization. Using the ordinary vector quantization method can solve the Context quantization, so clustering algorithm can be used to the Context quantization. However, the number of clusters is given basically in the clustering algorithm. How many it makes the clustering effectively is the main problem. K-means is used widely, K-means algorithm needs to initialize the cluster centers and the number of clusters is given,so it is more sensitive to these. Therefore, this will have to face two problems, given the initial cluster centers randomly can or not get the best clustering results and how many the number of clusters is when the clustering results is the optimal. The genetic algorithm is a global optimization algorithm developed in recent years, it borrowed from the biological point of view of genetics, through natural selection, crossover and mutation mechanism, adaptive improvement of the individual. As long as in the choice of a suitable fitness function, you can get the best individuals in a population. The genetic algorithm is an extensive global search method and it has been applied in many fields.To find the best number of clusters and reduce the impact of clustering initialized cluster centers for the clusters, as genetic K-means clustering algorithm is used to the research of Context quantization. Genetic K-means is a combination of genetic algorithm and K-means algorithm Among them, genetic algorithms use the mechanism of natural selection to get the best individual, and thus, the effective allele in the Chromosome as the initialization of the K-means clustering center, and the number of effective genes in the chromosome as the number of class clustering. Therefore, solving the impact of the initial conditions to K-means, genetic algorithm is a global search of ways and overcome the shortcomings of the local optimal K-means faces. Therefore, the genetic K-means can find the optimal of Context quantizer. The experiments show that Genetic K-means algorithm is better than K-means algorithm based Context quantization, it can find the best number of clusters and get the best code length.