Augmentation Is What You Need!

Abstract

We investigate the effect of augmentation of SMILES to increase the performance of convolutional neural network models by extending the results of our previous study [1] to new methods and augmentation scenarios. We demonstrate that augmentation significantly increases performance and this effect is consistent across investigated methods. The convolutional neural network models developed with augmented data on average provided better performances compared to those developed using calculated molecular descriptors for both regression and classification tasks.