machine learning - Implementation of SVM for classification without library in c++ -

i'm studying support vector machine last few weeks. understand theoretical concept how can classify data 2 classes. unclear me how select support vector , generate separating line classify new data using c++.

suppose, have 2 training data set 2 classes

after plotting data, following feature space vector , here, separating line clear.

how implement in c++ without library functions. me clear implementation concept svm. need clear implementation i'm going apply svm in opinion mining native language.

i join people's advice , should consider using library. svm algorithm tricky enough add noise if not working because of bug in implementation. not talking how hard make scalable implementation in both memory size , time.

that said , if want explore learning experience, smo best bet. here resources use:

the simplified smo algorithm - stanford material pdf

fast training of support vector machines - pdf

the implementation of support vector machines using sequential minimal optimization algorithm - pdf

probably practical explanation have found 1 on chapter 6 of book machine learning in action peter harrington. code on python should able port c++. don't think best implementation might enough have idea of going on.

the code freely available:

https://github.com/pbharrin/machinelearninginaction/tree/master/ch06

unfortunately there not sample chapter lot of local libraries tend have book available.