Subspace learning based classification of ADHD patients

Attention deficit hyperactivity disorder (ADHD) can seriously affect children's development. Effective diagnosis of ADHD is widely concerned, but accurate diagnosis is still a challenge. In this paper, we propose a classification method based on the subspace learning. Specifically, the corresponding control subspace and patient subspace are learned according to subspace metrics of the control group and the patient group, and linear regression constraints are used to learn the most important and distinctive information. In the learned subspace, subjects have greater energy in the corresponding subspace. Compared with several advanced machine learning methods, our method shows significant classification performance, with an accuracy rate of 94.6% in the ADHD-200 database.