Implementing Support Vector Machine (SVM) usng LIBSVM in MATLAB

Hi folks,

My first post concerns an exiting project I am working with, relating to implementing the SVM on time series data using LIBSVM (code available via LIBSVM -- A Library for Support Vector Machines) in MATLAB.

Though I understand a bit of how SVM works mathematically (I have only taken one course in applied linear optimization, so my competence in this field is limited), I have no idea how to implement it. I have read all the popular tutorials on the subject, alas, they assume you know the basics. The basics include:

1. Choosing features
2. choosing classes

using MATLAB code.
My main question relates to what do to with my data (ex. how to group it into features in order to use the cross-validation function, how to adapt the data to use the svm-train function etc).

Has someone here implmented SVM using MATLAB, and has the time and kindness to help me out?

Hmm, let me be more precise. Ex. say I have 1000 samples, and I want to check whether they have certain features.
The samples are not of the same length however, so putting them into a matrix is not easy without tampering with the samples.
Yet th esupport vector methods require one to work with matrixes. Should I simply test one sample at a time for the features?