|Table of Contents|

Incremental learning algorithm based on selforganizing map and probabilistic neural network


Research Field:
Publishing date:


Incremental learning algorithm based on selforganizing map and probabilistic neural network
Qi Yong12Hu Jun1Yu Dongjun12
1.School of Computer Science and Engineering,NUST,Nanjing 210094,China; 2.Changshu Institute,NUST,Changshu 215513,China
selforganizing mapprobabilistic neural networkincremental learningmachine learning
To solve the defects of the traditional learning algorithms that the new available data can not be effectively utilized,a new incremental learning method called incremental modular selforganizing map probabilistic neural network(IMSOMPNN),based on selforganizing map(SOM)and probabilistic neural network(PNN)is proposed.Samples of each class are used to train a modular SOM and the codebook vectors of the trained SOMs are used as pattern neurons for constructing PNN.The proposed IMSOMPNN possesses several advantages such as:(1)it can easily learn the knowledge buried in different types of new available data;(2)only the new available data are used to update the trained model,and the original data do not need;(3)the proposed IMSOMPNN has good performance even on noisy data.Experimental results on the UCI Landsat Satellite dataset demonstrate the effectiveness of the proposed method.


[1]He H B,Chen S,Li K,et al.Incremental learning from stream data[J].IEEE Transactions on Neural Networks,2011,22(12):1901-1914.
[2]Wang Z L,Jiang M,Hu Y H,et al.An incremental learning method based on probabilistic neural networks and adjustable fuzzy clustering for human activity recognition by using wearable sensors[J].IEEE Transactions on Information Technology in Biomedicine,2012,16(4):691-699.
[3]Huang D,Yi Z,Pu X R.A new incremental PCA algorithm with application to visual learning and recognition[J].Neural Processing Letters,2009,30(3):171-185
[4]Zhao H,Yuen P C.Incremental linear discriminant analysis for face recognition[J].IEEE Trans Syst Man Cybern B Cybern,2008,38(1):210-221.
[5]Kohonen T.The selforganizing map[J].Proceedings of IEEE,1990,78(1):1461-1480.
[6]Specht D F.Probabilistic neural networks and the polynomial Adaline as complementary techniques for classification[J].IEEE Transactions on Neural Networks,1990,1(1):111-121.
[7]Ma F,Wang W P,Tsang W W,et al.Probabilistic segmentation of volume data for visualization using SOMPNN classifier[A].Proceedings of the 1998 IEEE Symposium on Volume Visualization[C].North Carolina,United States:ACM,1998:121-130.
[8]Yu D J,Shen H B,Yang J Y.SOMPNN:An efficient nonparametric model for predicting transmembrane helices[J].Amino Acids,2012,42(6):2195-2205.
[9]於东军,谌贻华,于海瑛.融合SOM与WangMendel方法的模糊规则提取[J].南京理工大学学报,2011,36(6):759-763. Yu Dognjun,Chen Yihua,Yu Haiying.Fuzzy rule extraction by fusing SOM and WangMendel method[J].Journal of Nanjing University of Science and Technology,2011,36(6):759-763.


Last Update: 2013-02-15