Machine Learning for Chemistry
This is my starting github repository for using TensorFlow in order to perform machine learning for computational chemistry.
So far, this is quite bare.
This example is based on the work of Steven Kearnes, et al. in Molecular graph convolutions: moving beyond fingerprints. In it, I create a dataset of the solubility of approximately 1000 chemicals in the same format that they used in identifying molecular features, and trained it briefly using a deep neural network and TensorFlow. The results are shown in Figure_1.png and while they are not nearly as good as Kearnes' model it was useful as a first project.