Machine Learning for Accelerated Molecular Simulation and Discovery

  • Date:

    July 12

  • Speaker:

    Pascal Friederich (KIT, Germany)

  • Time:

    15:30 - 16:05

  • Machine learning can couple different computational methods to enable and accelerate the development of new molecules and materials in multiple ways, e.g. by learning from large amounts of data to predict molecular or materials properties faster and screen large amounts of molecules, or by using optimization methods such as reinforcement learning directly for molecular design tasks. In this talk, I will give a brief overview of our research activities on graph neural networks for materials property prediction [1], machine learning accelerated atomistic simulations of photochemical reactions [2,3], and combining active learning and reinforcement learning to optimize molecular properties despite high computational costs for reward calculation [4].

     

    References

     

    [1] P. Reiser, A. Eberhard, P. Friederich, Software Impacts, 9, 100095 (2021)

     

    [2] P. Friederich, F. Häse, J. Proppe, A. Aspuru-Guzik, Nat. Mater., 20, 750-761 (2021)

     

    [3] J. Li, P. Reiser, B. Boswell, A. Eberhard, N. Burns, P. Friederich, S. Lopez, Chem. Sci., 12, 5302-5314 (2021)

     

    [4] A. Eberhard, H. Metni, P. Friederich, in preparation.