Relevance of Rotationally Equivariant Convolutions for Predicting Molecular Properties

Abstract

Equivariant neural networks (ENNs) are graph neural networks embedded in ${\mathbb{R}}^3$ and are well suited for predicting molecular properties. The ENN library e3nn has customizable convolutions, which can be designed to depend only on distances between points, or also on angular features, making them rotationally invariant, or equivariant, respectively. This paper studies the practical value of including angular dependencies for molecular property prediction directly via an ablation study with \texttt{e3nn} and the QM9 data set. We find that, for fixed network depth and parameter count, adding angular features decreased test error by an average of 23%. Meanwhile, increasing network depth decreased test error by only 4% on average, implying that rotationally equivariant layers are comparatively parameter efficient. We present an explanation of the accuracy improvement on the dipole moment, the target which benefited most from the introduction of angular features.

arXiv commentMachine Learning for Molecules Workshop at NeurIPS 2020, NeurIPS workshop on Interpretable Inductive Biases and Physically Structured Learning

References in corpus (14)

• PyTorch: An Imperative Style, High-Performance Deep Learning Library
• Gaussian Approximation Potentials: the accuracy of quantum mechanics, without the electrons
• Fast and Accurate Modeling of Molecular Atomization Energies with Machine Learning
• SchNet - a deep learning architecture for molecules and materials
• ANI-1: An extensible neural network potential with DFT accuracy at force field computational cost
• Machine Learning Unifies the Modelling of Materials and Molecules
• Don't Decay the Learning Rate, Increase the Batch Size
• Tensor field networks: Rotation- and translation-equivariant neural networks for 3D point clouds
• SchNetPack: A Deep Learning Toolbox For Atomistic Systems
• Train longer, generalize better: closing the generalization gap in large batch training of neural networks
• Revisiting Small Batch Training for Deep Neural Networks
• 3D Steerable CNNs: Learning Rotationally Equivariant Features in Volumetric Data
• Ensemble Learning of Coarse-Grained Molecular Dynamics Force Fields with a Kernel Approach
• Geometric Prediction: Moving Beyond Scalars

Cited by in corpus (17)

• The Open Catalyst 2020 (OC20) Dataset and Community Challenges
• Machine Learning on Neutron and X-Ray Scattering
• E(n) Equivariant Graph Neural Networks
• Rotation Invariant Graph Neural Networks using Spin Convolutions
• Predicting tensorial molecular properties with equivariant machine-learning models
• Learning Gradient Fields for Molecular Conformation Generation
• LieTransformer: Equivariant self-attention for Lie Groups
• Temporally coherent backmapping of molecular trajectories from coarse-grained to atomistic resolution
• Optimal radial basis for density-based atomic representations
• An Introduction to Electrocatalyst Design using Machine Learning for Renewable Energy Storage
• Mean-Field Density Matrix Decompositions
• Geometric and Physical Quantities Improve E(3) Equivariant Message Passing
• Machine learning and invariant theory
• Coordinate Independent Convolutional Networks -- Isometry and Gauge Equivariant Convolutions on Riemannian Manifolds
• Symmetry-Aware Actor-Critic for 3D Molecular Design
• Flexible dual-branched message passing neural network for quantum mechanical property prediction with molecular conformation
• Learning Atomic Multipoles: Prediction of the Electrostatic Potential with Equivariant Graph Neural Networks