Here is my problem: I am developing an embedded system for some classification task. I am using Logistic Regression as my classifier. Now I train my classifier, and download my model on to my machine. Now this is not everything I want to do. I now want to let my model adjust its classification boundary on runtime, online learning as it is called. How do I do it?