AINet-SL: Artificial immune network with social learning and its application in FIR filter designing

• An artificial immune network with social learning (AINet-SL) for optimization is proposed.
• Two social learning mechanisms, i.e., stochastic social learning (SSL) and heuristic social learning (HSL), are proposed.
• The numerical simulation results confirm the superiority of the proposed AINet-SL in solution accuracy and convergence speed.

This paper proposes an artificial immune network with social learning (AINet-SL) for complex optimization problems. In AINet-SL, antibodies are divided into two swarms. One is an elitist swarm (ES) where antibodies experience self-learning and the other is a common swarm (CS) where antibodies experience social-learning with different mechanisms, i.e., stochastic social-learning (SSL) and heuristic social-learning (HSL). The elitist antibody to be learned is selected randomly in SSL, while it is determined by the affinity measure in HSL. In order to obtain more accurate solutions, a dynamic searching step length updating strategy is proposed. A series of comparative numerical simulations are arranged among the proposed AINet-SL optimization, Differential Evolution (DE), opt-aiNet, IA-AIS and AAIS-2S. Five benchmark functions and a practical application of finite impulse response (FIR) filter designing are selected as testbeds. The simulation results indicate that the proposed AINet-SL is an efficient method and outperforms DE, opt-aiNet, IA-AIS and AAIS-2S in convergence speed and solution accuracy.

Figure optionsDownload as PowerPoint slide