Pre-training molecular graph representation with 3D geometry: rethinking self-supervised learning on structured data

Par Conseil national de recherches du Canada

Téléchargement	Voir la version finale : Pre-training molecular graph representation with 3D geometry: rethinking self-supervised learning on structured data (PDF, 6.3 Mio)
Auteur	Rechercher : Liu, Shengchao; Rechercher : Wang, Hanchen; Rechercher : Liu, Weiyang; Rechercher : Lasenby, Joan; Rechercher : Guo, Hongyu¹; Rechercher : Tang, Jian
Affiliation	Conseil national de recherches du Canada. Technologies numériques
Format	Texte, Article
Conférence	Thirty-fifth Conference on Neural Information Processing Systems (NeurIPS 2021), December 6-14, 2021, Virtual-only
Description physique	19 p.
Résumé	Molecular graph representation learning is a fundamental problem in modern drug and material discovery. Molecular graphs are typically modeled by their 2D topological structures, but it has been recently discovered that 3D geometric information plays a more vital role in predicting molecular functionalities. However, the lack of 3D information in real-world scenarios has significantly impeded the learning of geometric graph representation. To cope with this challenge, we propose the Graph Multi-View Pre-training (GraphMVP) framework where self-supervised learning (SSL) is performed by leveraging the correspondence and consistency between 2D topological structures and 3D geometric views. GraphMVP effectively learns a 2D molecular graph encoder that is enhanced by richer and more discriminative 3D geometry. We further provide theoretical insights to justify the effectiveness of GraphMVP. Finally, comprehensive experiments show that GraphMVP can consistently outperform existing graph SSL methods.
Date de publication	2021-12
Maison d’édition	Conference on Neural Information Processing Systems
Dans	Thirty-fifth Conference on Neural Information Processing Systems (NeurIPS 2021). NeurIPS 2021 Workshop: Self-Supervised Learning: Theory and Practice (décembre 2021). https://sslneurips21.github.io/pages/Accepted%20Paper.html.
Langue	anglais
Publications évaluées par des pairs	Oui
Exporter la notice	Exporter en format RIS
Signaler une correction	Signaler une correction (s'ouvre dans un nouvel onglet)
Identificateur de l’enregistrement	7a567360-9027-4d4d-82d6-70950cc0014e
Enregistrement créé	2022-05-16
Enregistrement modifié	2022-05-17

Date de modification :: 2024-07-03