The library can be used to train an SVM-based classifier from a set of training examples. Each training example is a pair where the CAR is a n-dimensional vector of real values, and the CDR is 1 or -1 to indicate positive or negative training example, respectively. Given a set of training examples, CL-SVM will output a function that takes as input a vector and returns as output T or NIL depending on whether it believes the input vector is a positive or negative example.
CL-SVM is part of the Suave project. It can be found at http://common-lisp.net/project/suave/.
You can currently get CL-SVM from its own git repository:
git clone http://common-lisp.net/project/suave/git/cl-svm/.git
An alternative to cl-svm is cl-libsvm, which is a wrapper around a C library.
Topics: machine learning AI