An optimise ELM by league championship algorithm based on food images

Abstract

This paper presents an optimisation of extreme learning machine by league championship algorithm based on food images. Extreme Learning Machine (ELM) is an effective classifier because of the performance which is higher than other classifiers’ aspects. However, some important drawbacks still work as a hindrance like failure of optimal selection weights for the weights of the input-hidden layer and the output of the threshold. In spite of the wide number of problem-solving attempts, there was no solution to be considered effective. This paper presents the approach of hybrid learning and the League Championship Algorithm is used by for the purpose of selecting the input weights and the thresholds outputs. The experimental outcomes showed that the performance of proposed technique is superior as compared according to different scenarios of the measures to benchmark. The proposed method has achieved an overall accuracy of 95% for UEC food 100 dataset and 94% for UEC food 256 dataset comparing with 94% and 80% for baseline approaches.