Adaboost Feature selection in Attention Network task experiment

Background and Aim: One of the most valuable resources of human cognition is attention and it is necessary for learning. There are several elements involved for either improving or degrading of attention studied to evaluate the attention. In this study, Attention Network Task is used. This model has three different networks including alerting, orienting and executive control. Methods: A total sample of ۶۳ participants of both genders, ages ranging between ۲۰ and ۴۰ was used. Number of languages, ages, sport duration, education level, gender, forward n-back test, Mindful Attention Awareness Scale score, Positive and Negative Affect Schedule score are gathered. Adaptive boosting (Adaboost) regression model is developed in order to find the important features. Results: This paper aims to find the aspect of attention by different feature selection models for these networks and evaluate the features that are more relevant for identifying the attention. Conclusion: The more important feature in the fitted model are selected. Top three features are number of languages, n-back forward test and negative affect score.

Keywords:

Adaboost , Attention Network Task , Feature selection , regression