An adaptive learning model based genetic for facial expression recognition

The genetic algorithms (GAs) are the evolutionary learning process, which are applied to complex optimization problems to find the optimum results from other results. The GAs to achieve the proper results in the iteration process selects the fitter solutions from a solution space. In this process, the parent selections, crossover and mutation operations have the important roles to find the optimum results. In this paper, the honey bees mating process has been modeled to modify the GAs learning process. For the purpose of illustration, the learning process with the proposed GA, the fuzzy membership functions were tuned with the proposed GA. In the proposed hybrid model, the core of expression recognition system is the fuzzy rule based system to classify the facial expressions recognition. Therefore, the proposed genetic algorithm called queen bee algorithm (QBA) is used with the purpose of making better performance in learning process to improve the accuracy and robustness of the system. To evaluate the system performance, images from Cohn-Kanade database were used to obtain the best functions parameters. Results showed that the membership functions under the training process have tuned properly while the accuracy of classification with optimized parameters illustrated the rate of 98% in the train process.