Binary Chaotic Jaya Optimization for Cognitive State Assessment

Abstract

Cognitive State Assessment has a significant role in analyzing the mental status of personals involved in high-risk tasks where decision-making is important. In this paper, authors have proposed a model to classify the cognitive states accurately. In the model, subband statistical wavelet-based features are extracted. Every feature may not be important for the classification of cognitive workload and introduces the problem of higher dimensionality. To solve the problem of high dimensionality, Chaotic Jaya Optimization based binary feature selection model is proposed. The model has been designed such that it not only improves the classification accuracy but also selects the relevant features. The extensive experiment has been performed using different techniques, and results show that without feature selection, 73.3% maximum accuracy is obtained using decision tree classifier. Further optimization techniques are employed for feature selection, and results are improved up to 96.11%. The results are also compared with the existing techniques and it has been observed that the proposed approach gives maximum classification accuracy and converges at least number of iterations. In the proposed approach, features are also reduced up to its 60%.