Neural Message Passing for Quantum Chemistry
Нейронная передача сообщений для квантовой химии
2017-04-04
SCID: 54.1/s57x2tss
Discuss with AI
Message Passing Neural Networks (MPNNs)molecular property predictionmolecular symmetry invarianceneural message passingquantum chemistry benchmarks
Figures from the paper
Abstract (AI)
Supervised learning on molecules has incredible potential to be useful in chemistry, drug discovery, and materials science. Luckily, several promising and closely related neural network models invariant to molecular symmetries have already been described in the literature. These models learn a message passing algorithm and aggregation procedure to compute a function of their entire input graph. At this point, the next step is to find a particularly effective variant of this general approach and apply it to chemical prediction benchmarks until we either solve them or reach the limits of the approach. In this paper, we reformulate existing models into a single common framework we call Message Passing Neural Networks (MPNNs) and explore additional novel variations within this framework. Using MPNNs we demonstrate state of the art results on an important molecular property prediction benchmark; these results are strong enough that we believe future work should focus on datasets with larger molecules or more accurate ground truth labels.
Key Findings
1
MPNNs achieve state-of-the-art results on an important molecular property prediction benchmark.
2
The achieved results are strong enough that further progress requires datasets with larger molecules or more accurate ground truth labels.
3
The authors explore novel variations within the MPNN framework to improve molecular property prediction.
4
The paper unifies several molecular graph neural network models into a single Message Passing Neural Network (MPNN) framework.
Research Object
Molecules represented as graphs for supervised molecular property prediction
Research Subject
Message Passing Neural Networks (MPNNs) and their variants for learning/inference of molecular properties via learned message-passing and aggregation on molecular graphs, evaluated on chemical prediction benchmarks
Publication Details
Publication Date
2017-04-04
Journal
Publisher
ISSN
Cited by
5055
Open access PDF
Access Type
Author Information
Download PDF
Subscribe to digest
Cited by20
A Comprehensive Survey on Graph Neural Networks2020
Graph neural networks: A review of methods and applications2020
Dynamic Graph CNN for Learning on Point Clouds2019
MoleculeNet: a benchmark for molecular machine learning2017
Recent advances and applications of machine learning in solid-state materials science2019
Heterogeneous Graph Transformer2020
Graph convolutional networks: a comprehensive review2019
The rise of deep learning in drug discovery2018
Deep Learning on Graphs: A Survey2020
Graph Neural Networks: A Review of Methods and Applications2018
Machine learning in materials science2019
DScribe: Library of descriptors for machine learning in materials science2019
Accelerating the discovery of materials for clean energy in the era of smart automation2018
Molecular representations in AI-driven drug discovery: a review and practical guide2020
A Survey on Graph Neural Networks for Time Series: Forecasting, Classification, Imputation, and Anomaly Detection2024
Artificial intelligence in drug discovery: recent advances and future perspectives2021
When physics meets machine learning: a survey of physics-informed machine learning2025
Graph Neural Networks and Their Current Applications in Bioinformatics2021
Computing Graph Neural Networks: A Survey from Algorithms to Accelerators2021
Graph Neural Networks for Intrusion Detection: A Survey2023