DeepChem 1.1.0 Release Notes

Over the past three months I have been working on DeepChem at the Pande Lab at Stanford. Deepchem is a high quality open-source toolchain for deep-learning in drug discovery, materials science, and quantum chemistry. During that time I developed TensorGraph, a new Deep Learning model framework inspired by Keras. TensorGraph models networks as DAGs of computation. This makes it easier to create complex non linear network topologies. Current models of MoleculeNet are being ported over into the new framework.

For the next release we are targeting

  • Increased image support
  • Example reinforcement learning implementations (A3C)
  • GAN support

It is a very exciting time to be doing machine learning research in drug discovery. Check out the work on github or just download the docker image.