Title :
Joint investigation of cases using self-organized map network
Author :
Lihong, Mo ; Mingguang, Wang ; Jun, Ji
Author_Institution :
Fac. of Electron. & Electr. Eng., Huaiyin Inst. of Technol., Huai´´an, China
Abstract :
In order to overcome the manual mode of cases´ joint investigation, improve the objectivity of joint investigation , and provide further alarm information, a modified self-organized map network was selected to realize joint investigation. The modifications were done mainly on the initial weight vector, the learnong rate, competition layer nodes´ selection, and the neighborhood range. After simulating on 25 groups of 8 dimensional case data, 15 groups of them were divided into 7 classes. There were still other cases that had similar outputs, which were used to broaden investigation ideas and ascertain trial strategy. The results prove that the SOM network is effetive in joint investigation application, it can be applied in every kinds of case detection occasions.
Keywords :
learning (artificial intelligence); self-organising feature maps; SOM network; competition layer node selection; learning rate; self-organized map network; trial strategy; weight vector; Communities; Joints; Planning; Stress; Support vector machine classification; Training; Vectors; joint investigation; learning rate; self-organized map; simulation; termination conditions;
Conference_Titel :
Electronics, Communications and Control (ICECC), 2011 International Conference on
Conference_Location :
Zhejiang
Print_ISBN :
978-1-4577-0320-1
DOI :
10.1109/ICECC.2011.6066741
