Amphetamine-type stimulants (ATS) drug classification using shallow one-dimensional convolutional neural network

• Main Authors: Draman @ Muda, Azah Kamilah, Mohd Yusof, Norfadzlia, Pratama, Satrya Fajri, Carbo‑Dorca, Ramon
• Format: Article
• Language: English
• Published: Springer 2021
• Online Access: http://eprints.utem.edu.my/id/eprint/26828/2/MOLECULAR_DIVERSITY_2021_PAPER_1.PDF; http://eprints.utem.edu.my/id/eprint/26828/; https://link.springer.com/article/10.1007/s11030-021-10289-1; https://doi.org/10.1007/s11030-021-10289-1
• Institution: Universiti Teknikal Malaysia Melaka

Amphetamine-type stimulants (ATS) drug analysis and identifcation are challenging and critical nowadays with the emergence production of new synthetic ATS drugs with sophisticated design compounds. In the present study, we proposed a one-dimensional convolutional neural network (1DCNN) model to perform ATS drug classifcation as an alternative method. We investigate as well as explore the classifcation behavior of 1DCNN with the utilization of the existing novel 3D molecular descriptors as ATS drugs representation to become the model input. The proposed 1DCNN model is composed of one convolutional layer to reduce the model complexity. Besides, pooling operation that is a standard part of traditional CNN is not applied in this architecture to have more features in the classifcation phase. The dropout regularization technique is employed to improve model generalization. Experiments were conducted to fnd the optimal values for three dominant hyperparameters of the 1DCNN model which are the flter size, transfer function, and batch size. Our fndings found that kernel size 11, exponential linear unit (ELU) transfer function and batch size 32 are optimal for the 1DCNN model. A comparison with several machine learning classifers has shown that our proposed 1DCNN has achieved comparable performance with the Random Forest classifer and competitive performance with the others.