This paper presents improvements to SOSDM based on ideas gleaned from the Danger Theory of immunology. In the new model, antibodies emit a signal describing their current level of contentment -monitoring the total level of contentment in the system provides a mechanism for determining when an immune response should occur, i.e. when new antibodies should be produced. It also provides a method of detecting catastrophic changes in the environment, i.e. significant changes in input data, and thus provides a means of removing antibodies. The new system, dSOSDM, is shown to be more robust and better able to deal with dynamically changing databases than SOSDM.
展开▼
