Complementary dual detectors for effective classification

Abstract

In this paper we introduce a method of using a pair of complementary negative detectors. When both self and non-self antigens are given, we can build a pair of complementary negative detectors using self and non-self antigens respectively and augment the results given by the detectors. When self or non-self antigens change over time, antibodies of a negative detector that gives a false positive error for the change, are used to fill the holes of the other negative detector giving a false negative error. They try to adapt to the change in complementary ways.