An unsupervised vegetation classification algorithm based immune

A novel artificial immune-based algorithm in predicting forest cover types with cartographic variables, referred to as POOTAI, is presented. Firstly, the definition of immune cell, antibody, and antigen are given. Then, the dynamic models of immune response, immune regulation and immune memory are evolved, and the corresponding equations are established. Finally, it is tested by the well-known forest cover types data set of UCI (University of California at Irvine) and compared with other known algorithms. POOTAI shows that the classification accuracy is increased to 90.17%, which is higher than other classification algorithms. It has some good features such as continuous learning, dynamic adjustment, characteristics memory, and etc.