An Agent Approach to Immunity-based Complex Systems

点击翻译按钮获取中文摘要

The immune system, viewed as a complex system, has autonomous and adaptive agents (corresponding to many types of the immune cell). The agent is the main primitive to implement "immunity-based systems" on a computer or on computer networks. The "immunity-based system" is an information-processing paradigm that collects processing principles, which are peculiar in the immune system. Among the principles, we focused on the followings: (1) Distributed and network-based processing by agents, (2) "Self-nonself discrimination" by the self (nonself) reference, (3) Adaptive strategy by diversity. The research results can be summarized by the following four items.1. General Remarks (Part I of the report):A design paradigm to implement immunity-based system" as information-processing system is discussed. Models for simulating the immune system from the viewpoint of complex system are also discussed.2. Applications 1 (Part II of the report): The design paradigm above is further clarified as a specification. Applications to sensor-based diagnosis, adaptive disturbance rejection, and credibility evaluation of information gathering agents are presented.3. Applications 2 (Part III of the report): A prototype of an anti-virus system using a computer network against computer viruses is implemented. Simulations are conducted by viewing the spreading phenomena of computer viruses from the complex system view. Effective prevention against spreading is also considered.4. Theoretical Considerations (Part IV of the report): Diversity; the strategy used by the immune system is discussed by focusing on the interaction between HIV and the immune system. A dynamical model for the interaction is analyzed by a diversity measure.