Share your thoughts, 1 month free Claude Pro on usSee more
WorkDL logo mark

GraphDF: A Discrete Flow Model for Molecular Graph Generation

About

We consider the problem of molecular graph generation using deep models. While graphs are discrete, most existing methods use continuous latent variables, resulting in inaccurate modeling of discrete graph structures. In this work, we propose GraphDF, a novel discrete latent variable model for molecular graph generation based on normalizing flow methods. GraphDF uses invertible modulo shift transforms to map discrete latent variables to graph nodes and edges. We show that the use of discrete latent variables reduces computational costs and eliminates the negative effect of dequantization. Comprehensive experimental results show that GraphDF outperforms prior methods on random generation, property optimization, and constrained optimization tasks.

Youzhi Luo, Keqiang Yan, Shuiwang Ji• 2021

Related benchmarks

TaskDatasetResultRank
Molecular GenerationZINC250K
Uniqueness99.97
68
Graph generationENZYMES
Clustering1.061
45
Molecular GenerationQM9
Validity89
40
Molecular Graph GenerationQM9
Validity93.88
37
Property optimizationZINC250k (test)
1st Order Metric0.844
33
Molecular GenerationQM9 (test)
Validity93.88
32
Molecule GenerationZINC250K
Validity89.72
32
Constrained Property OptimizationZINC250K
Improvement5.62
27
Abstract graph generationEgo small
Average MMD0.06
27
Molecular GenerationZINC250k (test)
Validity90.61
26
Showing 10 of 25 rows

Other info

Follow for update