Timezone: »

 
Poster
Learning Invariant Representations of Molecules for Atomization Energy Prediction
Grégoire Montavon · Katja Hansen · Siamac Fazli · Matthias Rupp · Franziska Biegler · Andreas Ziehe · Alexandre Tkatchenko · Anatole von Lilienfeld · Klaus-Robert Müller

Wed Dec 05 07:00 PM -- 12:00 AM (PST) @ Harrah’s Special Events Center 2nd Floor #None

The accurate prediction of molecular energetics in chemical compound space is a crucial ingredient for rational compound design. The inherently graph-like, non-vectorial nature of molecular data gives rise to a unique and difficult machine learning problem. In this paper, we adopt a learning-from-scratch approach where quantum-mechanical molecular energies are predicted directly from the raw molecular geometry. The study suggests a benefit from setting flexible priors and enforcing invariance stochastically rather than structurally. Our results improve the state-of-the-art by a factor of almost three, bringing statistical methods one step closer to the holy grail of ''chemical accuracy''.

Author Information

Grégoire Montavon (TU Berlin)
Katja Hansen (Fritz-Haber-Institut)
Siamac Fazli (TU Berlin)
Matthias Rupp (ETH Zurich)
Franziska Biegler (TU Berlin)
Andreas Ziehe (Fraunhofer FIRST)
Alexandre Tkatchenko (University of Luxembourg)
Anatole von Lilienfeld (Argonne National Laboratory)
Klaus-Robert Müller (TU Berlin)

More from the Same Authors