Speaker: Dr. Kiyoshi Hasegawa,                                                                                                                                                                                                    Chugai Pharmaceutical Co., LTD.,  Division Chair of Chemoinformatics, The Chemical Society of Japan,

Title:  2D and 3D convolutional neural networks for molecular design

Abstract:

Deep learning such as convolutional neural networks (CNN) is powerful method for building the structure-activity relationship. CNN is useful for predictions of activities of new molecules. However, if we employ complicated model, we cannot make deep insights for the predictive model. This aspect is import for molecular design. In my talk, I have tried to interpret the 2D and 3D convolutional neural networks with chemical language.