The Research Of Negative Selection Algorithm And Visualization Implementation

Negative selection algorithm is a main technology of artificial immune algorithm. As a binary classification problem solving classification algorithm, it successfully simulated the immune system to recognize self and non-self immune tolerance process. In the study of algorithms simultaneously, data mining visualization tools have sprung up:Weka, R, SPSS, MatLab and other stand-alone version of the tool has been known to us all. In order to solve the problem which the data is too huge to handle by stand-alone tools, some analysis-platform based on big data in these years rapid development. However, in the most basic algorithm research area, there have not the relevant professional tools.The paper research on negative selection algorithm and visualization data classification platform.The main work is as follows:1) Proposed a new real-valued negative selection algorithm. It considered non-self information and the twice moved strategy, so we call it NTMV-detector. Firstly, proposed a new method of generating candidate-detector center, it gave priority to non-self information of training data. Secondly, in the process of rejecting or accepting candidate-detector center and join it into mature-detector set, adopting the twice move strategy, it improved the efficiency of the mature detector generation. The algorithm made full use of non-self information, and improved the mechanism of the detector. Without increasing the number of detector, it improved the detection efficiency.2) Designed a Platform of Visualization Data Classification (PVDC) solution and achieved its prototype system. When use PVDC, we could share our data and research result of algorithm to others, we also could compare and analyses the experiment result of different algorithm. It’s an integrated experiment platform for researchers of classification-algorithm.3) Use two of disease diagnosis data sets (BCW, Pima) experiment on the PVDC platform, the results were analyzed and compared with other similar algorithms. The experiment result expresses that NTMV-detector compared with other similar algorithms in dealing with disease diagnosis aspect can fairly effective improved diagnosis and reduced misdiagnosis rate. At the same time, verified PVDC is usable system.