AN OPTIMIZED ADABOOST ALGORITHM BASED ON K-MEANS CLUSTERING
Peng Zhang · Journal of Physics Conference Series · 2021
Abstract Classification plays an important role on data mining techniques. AdaBoost is a classic upgrade algorithm on data classification. The Optimization algorithm on AdaBoost emerged is endless. In this paper we make some improvement on the base of a nonlinear AdaBoost algorithm based on statistics for K-nearest neighbors. In the basic algorithm, the prediction accuracies were improved more or less while it took plenty of time to calculate the Euclidean distance between the instance and all the samples. So we raise an improved method to shorten the computing time with K-means Clustering algorithm, and the classification accuracies will be as accurate as the original. Experiment results show that improvement is an effective method while the number of samples is large. And the bigger the number of the samples is, the more time can be saving in the stage of classification for the instance.