%0 Journal Article %1 Smith2017ANAKIN %A Smith, J. S. %A Isayev, O. %A Roitberg, A. E. %D 2017 %I The Royal Society of Chemistry %J Chem. Sci. %K force-field machine-learning molecular-dynamics neural-network quantum-chemistry quantum-force-fields %N 4 %P 3192-3203 %R 10.1039/C6SC05720A %T ANI-1: an extensible neural network potential with DFT accuracy at force field computational cost %U http://dx.doi.org/10.1039/C6SC05720A %V 8 %X Deep learning is revolutionizing many areas of science and technology, especially image, text, and speech recognition. In this paper, we demonstrate how a deep neural network (NN) trained on quantum mechanical (QM) DFT calculations can learn an accurate and transferable potential for organic molecules. We introduce ANAKIN-ME (Accurate NeurAl networK engINe for Molecular Energies) or ANI for short. ANI is a new method designed with the intent of developing transferable neural network potentials that utilize a highly-modified version of the Behler and Parrinello symmetry functions to build single-atom atomic environment vectors (AEV) as a molecular representation. AEVs provide the ability to train neural networks to data that spans both configurational and conformational space, a feat not previously accomplished on this scale. We utilized ANI to build a potential called ANI-1, which was trained on a subset of the GDB databases with up to 8 heavy atoms in order to predict total energies for organic molecules containing four atom types: H, C, N, and O. To obtain an accelerated but physically relevant sampling of molecular potential surfaces, we also proposed a Normal Mode Sampling (NMS) method for generating molecular conformations. Through a series of case studies, we show that ANI-1 is chemically accurate compared to reference DFT calculations on much larger molecular systems (up to 54 atoms) than those included in the training data set.

@article{Smith2017ANAKIN, abstract = {Deep learning is revolutionizing many areas of science and technology{,} especially image{,} text{,} and speech recognition. In this paper{,} we demonstrate how a deep neural network (NN) trained on quantum mechanical (QM) DFT calculations can learn an accurate and transferable potential for organic molecules. We introduce ANAKIN-ME (Accurate NeurAl networK engINe for Molecular Energies) or ANI for short. ANI is a new method designed with the intent of developing transferable neural network potentials that utilize a highly-modified version of the Behler and Parrinello symmetry functions to build single-atom atomic environment vectors (AEV) as a molecular representation. AEVs provide the ability to train neural networks to data that spans both configurational and conformational space{,} a feat not previously accomplished on this scale. We utilized ANI to build a potential called ANI-1{,} which was trained on a subset of the GDB databases with up to 8 heavy atoms in order to predict total energies for organic molecules containing four atom types: H{,} C{,} N{,} and O. To obtain an accelerated but physically relevant sampling of molecular potential surfaces{,} we also proposed a Normal Mode Sampling (NMS) method for generating molecular conformations. Through a series of case studies{,} we show that ANI-1 is chemically accurate compared to reference DFT calculations on much larger molecular systems (up to 54 atoms) than those included in the training data set.}, added-at = {2018-11-18T20:03:14.000+0100}, author = {Smith, J. S. and Isayev, O. and Roitberg, A. E.}, biburl = {https://www.bibsonomy.org/bibtex/2304ee717a153e3b1a91ff85cac15470f/salotz}, description = {ANI-1: an extensible neural network potential with DFT accuracy at force field computational cost - Chemical Science (RSC Publishing) DOI:10.1039/C6SC05720A}, doi = {10.1039/C6SC05720A}, interhash = {519ea864c2d44df5a2e443da6cc2e810}, intrahash = {304ee717a153e3b1a91ff85cac15470f}, journal = {Chem. Sci.}, keywords = {force-field machine-learning molecular-dynamics neural-network quantum-chemistry quantum-force-fields}, number = 4, pages = {3192-3203}, publisher = {The Royal Society of Chemistry}, timestamp = {2018-11-18T20:03:14.000+0100}, title = {ANI-1: an extensible neural network potential with DFT accuracy at force field computational cost}, url = {http://dx.doi.org/10.1039/C6SC05720A}, volume = 8, year = 2017 }