Hi,

you can try:

minimisation of both classification error probability (POE) and average correlation coefficients (ACC) between

(M. Dash and H. Liu, Feature selection for Classification (1997) Elsevier Science Inc)

Mutual information measure

(G. D. Tourassi, E. D. Frederick, M. K. Markey and C. E. Floyd, Jr. (2001) Application of the mutual information criterion for feature selection in computer-aided diagnosis, Medical Physics, 28, 12, 2394-2402.)

Hi,

First you need more information about Feature selection. I suggest you to read this paper:

http://www.sciencedirect.com/science/article/pii/S0045790613003066

then you can study these papers to solve your problem:

http://repository.ias.ac.in/26053/

http://link.springer.com/article/10.1023/A:1010933404324

http://www.hindawi.com/journals/jam/2013/754698/abs/

http://link.springer.com/chapter/10.1007/978-3-642-35326-0_2

good luck.

we can use fuzzy logic or other classification method as SVM

Dear,

Try to check  ReliefF, where it  is considered stable with GLABRA,

TOX, and PRO-CAN..

can refer to:

http://nar.oxfordjournals.org/content/early/2012/01/18/nar.gkr1265.full.html

Feature Selection Library (FSLib 2018) is a widely applicable MATLAB library for feature selection.

Link: https://uk.mathworks.com/matlabcentral/fileexchange/56937-feature-selection-library

In the toolbox you can find 19 recent methods (some 2015-2017-2018):

[1] ILFS [2] InfFS [3] ECFS [4] mrmr [5] relieff [6] mutinffs [7] fsv [8] laplacian [9] mcfs [10] rfe [11] L0 [12] fisher [13] UDFS [14] llcfs [15] cfs [16] fsasl [17] dgufs [18] ufsol [19] lasso