Discovering the Characteristic Set of Metaheuristic Algorithm to Adapt with ANFIS Model
share
In recent years, many applications based on the neural network, neuro fuzzy and optimization algorithms have become increasingly common for solving regression and classification problems. In the adaptive neuro-fuzzy inference system (ANFIS), many researchers used the adaption of metaheuristic algorithms with ANFIS to propose the best estimation model. However, many researchers only focused on the experiment without a mathematical demonstration or indicating which characteristic of the optimization algorithm, during run, affect and settable, are in coordination with ANFIS. This paper provides an adaption of metaheuristic algorithms with ANFIS which has been performed by considering accuracy parameters in layer 1 and layer 4 for the estimation problem. It has integrated six well-known metaheuristic algorithms and extracted their characteristics. In the experiment, the metaheuristic algorithms based on the evolutionary computation have been demonstrated more stable than swarm intelligence methods in tuning parameters of ANFIS.
READ FULL TEXT
