The server is under maintenance between 08:00 to 12:00 (GMT+08:00), and please visit later.
We apologize for any inconvenience caused
Login  | Sign Up  |  Oriprobe Inc. Feed
China/Asia On Demand
Journal Articles
Laws/Policies/Regulations
Companies/Products
Bookmark and Share
Research on the K-D Tree KNN-SVM Classifier in Underwater Acoustic Target Recognition
Author(s): 
Pages: 15-22
Year: Issue:  1
Journal: Ocean Technology

Keyword:  underwater target recognitionsupport vector machine(SVM)K nearest neighbor(KNN)K-D treeKNN-SVM combined classifier;
Abstract: Conventional KNN-SVM classifier combination in K-nearest neighbor algorithm does not fully exploit the information of training samples. Researchers used to adopt the traversal method to calculate the distance between the recognition and training samples, leading to a large number of redundant computations, especially when training samples are great. Aiming at the problem of training the training samples into a K-D tree structure,this paper designs a K-D tree KNN-SVM classifier, which can greatly reduce the redundant calculation, so as to improve the searching efficiency, thus effectively shortening the search time. The simulation and experimental research are carried out, and the KNN, SVM and KNN-SVM classifiers are designed to classify the two kinds of underwater targets, with relevant parameters optimized. The experimental results show that after selecting the best parameters, the KNN-SVM classifier combination is the optimal method compared with the other two types of classifiers in the recognition rate and efficiency. The KNN part recognition efficiency is 7.5 times higher than conventional when adopting K-D tree structure of the KNN-SVM classifier combination.
Related Articles
No related articles found