|
|
| Type |
Poster Presentation |
| Area |
Medicinal Chemistry |
| Room No. |
Exhibition Hall 2+3 |
| Time |
10월 19일 (목요일) 11:00~12:30 |
| Code |
MEDI.P-327 |
| Subject |
QSAR Predictability Comparison between Deep Neural Network and Traditional Machine Learning Methods |
| Authors |
Yongil Seo, Young Ju Seo, YongJoon Jang, JA YEON MOON, Namseok Kim, Ky-Youb Nam*, Sinyoung Kim1, Kwang-Hwi Cho1, Jeong Hyeok YOON2
Bio Artificial Intelligence Research Center, Pharos I&BT Co., Ltd., Korea
1Department of Bioinformation, Soongsil University, Korea
2Pharos I&BT Co., Ltd., Korea
|
| Abstract |
In recent years, artificial neural network (ANN) model have been improved and the development of hardware has attracted the interest to deep neural network (DNN) and it has been applied in various fields example for automatic speech recognition, image recognition and customer relationship management. In particular, various machine learning techniques such as support vector machine (SVM) and multiple linear regression (MLR) as well as neural networks have been used in the field of quantitative structure-activity relationship (QSAR) for predicting the activities of compounds. In this study, we performed DNN methods to predict the toxicity of compounds by building some predicting models of hERG inhibitors which has an important toxicity on the discovery of drugs. In addition, we compared predictability between the DNN and the machine learning techniques such as SVM, MLR and ANN that have been previously developed. In the quantitative models, the MLR and ANN methods were compared with the DNN method. In the qualitative model, the SVM and Random forest methods were compared the DNN method. |
| E-mail |
tntmc@naver.com |
|
120th General Meeting of the KCS