Artificial Immune Systems: Part Ii – a Survey of Applications

Figure 1: (a) Environment for testing the consensus-making algorithm based on immune networks. Figure 2: Immunized computational system structure. BCS corresponds to the base-line computational system (representing an average behavior of the uncertain system), and CCS corresponds to the changeable computational system (representing the variable region of the antibody and epitope equivalents). Figure 3: Encoding scheme for the case of a Fuzzy and a Neural Network building block. The parameter 'variable' can assume any linguistic value (altitude, speed, etc.), and 'relationship' represents any logical antecedent (AND, OR, etc.) or consequent (THEN). The neural net building block is responsible for describing the network architecture and weight values. Figure 4: (a) Example of structural optimization problem studied. (b) Constraint elimination using immune principles. Figure 6: The clonal selection algorithm. (a) Block diagram. (b) Step-by-step procedure. Figure 7: (a) n-TSP agents. (b) Proposed immune algorithm (IA) for solving the n-TSP problem. Each immune cell set is composed of three kinds of cells, called a macrophage, a B and a T-cell. Figure 8: ABNET. (a) Main steps of the learning algorithm. (b) Weights updating procedure. Figure 9: Basic features of the network supervised learning algorithm without changing the synaptic connection strengths (W ij). When a set of stimuli has been learnt, the system remains in the cycle shown in double line arrows. Figure 10: Anomaly detection algorithm. (a) Generation of valid detector set (censoring). Figure 13: (a) Genetic encoding for the detectors. (b) Suppression of detector C that (partially) covers the same portion of the pattern space. Figure 18: (a) Schematic diagram of the immune algorithm. (b) Immune algorithm for an agent-based architecture. ...34 Figure 19: (a) The immune system object algorithm. (b) Structure of a B-cell object.. Figure 21: Non-linear relation between binding value and match score for a bit string of length L = 7 and affinity threshold ε = 2. Figure 24: (a) Rules for the proposed cellular automaton model. (b) Steps in the simulation. Figure 26: (a) Structure of an antibody in the iNet. (b) A UML class diagram for a kernel of iNet. Figure 28: Integer-valued encoding for the antigen and antibody molecules, together with the matching function. A match between number corresponds to a score of 5 and a don't care corresponds to a score of 1. Figure 30: (a) Structure of a case (antigen or antibody). (b) The match algorithm (see Equation (11)). Figure 32: (a) Typical …