Timezone: »

On the equivalence of molecular graph convolution and molecular wave function with poor basis set
Masashi Tsubaki · Teruyasu Mizoguchi

Wed Dec 09 09:00 AM -- 11:00 AM (PST) @ Poster Session 3 #964

In this study, we demonstrate that the linear combination of atomic orbitals (LCAO), an approximation introduced by Pauling and Lennard-Jones in the 1920s, corresponds to graph convolutional networks (GCNs) for molecules. However, GCNs involve unnecessary nonlinearity and deep architecture. We also verify that molecular GCNs are based on a poor basis function set compared with the standard one used in theoretical calculations or quantum chemical simulations. From these observations, we describe the quantum deep field (QDF), a machine learning (ML) model based on an underlying quantum physics, in particular the density functional theory (DFT). We believe that the QDF model can be easily understood because it can be regarded as a single linear layer GCN. Moreover, it uses two vanilla feedforward neural networks to learn an energy functional and a Hohenberg--Kohn map that have nonlinearities inherent in quantum physics and the DFT. For molecular energy prediction tasks, we demonstrated the viability of an ``extrapolation,'' in which we trained a QDF model with small molecules, tested it with large molecules, and achieved high extrapolation performance. We believe that we should move away from the competition of interpolation accuracy within benchmark datasets and evaluate ML models based on physics using an extrapolation setting; this will lead to reliable and practical applications, such as fast, large-scale molecular screening for discovering effective materials.

Author Information

Masashi Tsubaki (National Institute of Advanced Industrial Science and Technology (AIST))
Teruyasu Mizoguchi (University of Tokyo)

2002 PhD, Kyoto University, Japan 2002-2003 JSPS research fellow, Kyoto University, Japan 2003-2004 JSPS research fellow, University of Tokyo, Japan 2004-2005 JSPS research fellow, Lawrence Berkeley National Lab., USA 2005-2007 Research Assistant, University of Tokyo, Japan 2007-2009 Assistant Professor, University of Tokyo, Japan 2009-present Associate Professor, University of Tokyo, Japan

More from the Same Authors

  • 2018 Poster: Mean-field theory of graph neural networks in graph partitioning »
    Tatsuro Kawamoto · Masashi Tsubaki · Tomoyuki Obuchi
  • 2017 : Poster session 1 »
    Van-Doan Nguyen · Stephan Eismann · Haozhen Wu · Garrett Goh · Kristina Preuer · Thomas Unterthiner · Matthew Ragoza · Tien-Lam PHAM · Günter Klambauer · Andrea Rocchetto · Maxwell Hutchinson · Qian Yang · Rafael Gomez-Bombarelli · Sheshera Mysore · Brooke Husic · Ryan-Rhys Griffiths · Masashi Tsubaki · Emma Strubell · Philippe Schwaller · Théophile Gaudin · Michael Brenner · Li Li
  • 2017 : Poster spotlights »
    Emma Strubell · Garrett Goh · Masashi Tsubaki · Théophile Gaudin · Philippe Schwaller · Matthew Ragoza · Rafael Gomez-Bombarelli