Predicting molecular properties with covariant compositional networks.

By Truong Son Hy, Shubhendu Trivedi, Horace Pan, Brandon M. Anderson, Risi Kondor. 2018

73 citations recorded by the source corpus.

Density functional theory (DFT) is the most successful and widely used approach for computing the electronic structure of matter. However, for tasks involving large sets of candidate molecules, running DFT separately for every possible compound of interest is forbiddingly expensive. In this paper, we propose a neural network based machine learning algorithm which, assuming a sufficiently large training sample of actual DFT results, can instead learn to predict certain properties of molecules purely from their molecular graphs. Our algorithm is based on the recently proposed covariant compositional networks framework and involves tensor reduction operations that are covariant with respect to permutations of the atoms. This new approach avoids some of the representational limitations of other neural networks that are popular in learning from molecular graphs and yields promising results in numerical experiments on the Harvard Clean Energy Project and QM9 molecular datasets.

Subfields: Computational Theory and Mathematics, Materials Chemistry, Physical and Theoretical Chemistry, Graph Learning, Computational Chemistry & Drug Discovery

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Computational Theory and Mathematics

1. Max Welling
2. Maurice Weiler
3. Stephan Günnemann
4. Johannes Klicpera
5. Erik J. Bekkers
6. Taco Cohen
7. Janek Groß
8. Berkay Kicanaoglu

In Materials Chemistry

1. Max Welling
2. Stephan Günnemann
3. Johannes Klicpera
4. Maurice Weiler
5. Tess E. Smidt
6. Taco Cohen
7. Janek Groß
8. Yi Liao

In Physical and Theoretical Chemistry

1. Janek Groß
2. Max Welling
3. Stephan Günnemann
4. Johannes Klicpera

In Graph Learning

1. Stephan Günnemann
2. Johannes Klicpera
3. Max Welling
4. Maurice Weiler
5. Taco Cohen
6. Berkay Kicanaoglu
7. Janek Groß
8. William L. Hamilton

In Computational Chemistry & Drug Discovery

1. Max Welling
2. Stephan Günnemann
3. Johannes Klicpera
4. Janek Groß
5. Taco Cohen
6. Berkay Kicanaoglu
7. Maurice Weiler
8. Tess E. Smidt

Citations in the corpus

• Directional Message Passing for Molecular Graphs (2020, background): “ks, which do not require any hand-crafted features but learn representations solely based on the atom types and coordinates molecules (Duvenaud et al., 2015; Gilmer et al., 2017; Schütt et al., 2017; Hy et al., 2018; Unke & Meuwly, 2019). Our proposed message embeddings can also be interpreted a”
• Gauge Equivariant Convolutional Networks and the Icosahedral CNN (2019, background): “, 2018), graphs (Kondor et al., 2018b; Hy et al., 2018; Maron et al., 2019), translations and rotations of the plane and 3D space (Oyallon & Mallat, 2015; Cohen & Welling, 2016; 2017; Marcos et al.”