On The Incremental Clustering Algorithms Based On Kalman Filter

While the era of big data brings convenience to life,the rapid expansion of data also brings many challenges.For individuals,enterprises,governments and other parties,it is crucial to dig out valuable information in the existing big data.This involves the complexity of the data and how to get the most important information in a reasonable amount of time.The main research content of this paper is the incremental clustering algorithm based on Kalman Filter.It is of practical significance to discuss the improved algorithm principle and performance.The main research work of this paper is as follows:1.After understanding the research status of clustering algorithm at home and abroad,combining the clustering algorithm with the principle and method of Kalman Filter to calculate the mean value of data set,the effect is better.This method provides a new idea for incremental clustering algorithm.2.Apply Kalman Filter idea to clustering algorithm,propose two propositions for single class and multiple class incremental clustering algorithm,and conduct theoretical derivation;Meanwhile,two incremental clustering algorithms based on Kalman Filter are proposed: ISCC algorithm based on a single class and IMCC algorithm based on multiple classes.The above two incremental clustering algorithms are simulated by MATLAB,and the results show that ISCC algorithm and IMCC algorithm are feasible and effective.3.For highly complex data sets,incremental clustering algorithm may sometimes fail to achieve an ideal effect when only mean and variance are considered.In order to improve the accuracy of the algorithm,the paper introduces a semi-supervised incremental clustering algorithm.Two semi-supervised incremental clustering algorithms,KQDBHIC algorithm and KBBHIC algorithm,based on quadratic discriminant and Bayes discriminant respectively,are proposed.Comparing KQDBHIC algorithm and KBBHIC algorithm with the existing algorithm,the accuracy has been significantly improved and the effect is good,which proves that the algorithm constructed in this paper has certain advantages.