CTAGE: Curvature-Based Topology-Aware Graph Embedding for Learning Molecular Representations
Molecular graph
Graph Embedding
Topological graph theory
DOI:
10.48550/arxiv.2307.13275
Publication Date:
2023-01-01
AUTHORS (5)
ABSTRACT
AI-driven drug design relies significantly on predicting molecular properties, which is a complex task. In current approaches, the most commonly used feature representations for training deep neural network models are based SMILES and graphs. While these methods concise efficient, they have limitations in capturing spatial information. Recently, researchers recognized importance of incorporating three-dimensional information structures into models. However, requires introduction additional units generator, bringing computational costs. Therefore, it necessary to develop method properties that effectively combines structural while maintaining simplicity efficiency graph networks. this work, we propose an embedding approach CTAGE, utilizing $k$-hop discrete Ricci curvature extract insights from data. This integrates preserving complexity network. Experimental results indicate introducing node improves performance frameworks, validating k-hop reflects relationship between structure function.
SUPPLEMENTAL MATERIAL
Coming soon ....
REFERENCES ()
CITATIONS ()
EXTERNAL LINKS
PlumX Metrics
RECOMMENDATIONS
FAIR ASSESSMENT
Coming soon ....
JUPYTER LAB
Coming soon ....