DiGress: Discrete Denoising Diffusion for Graph Generation | Clément Vignac

Join the Learning on Graphs and Geometry Reading Group: https://hannes-stark.com/logag-readin...

Paper "DiGress: Discrete Denoising Diffusion for Graph Generation": https://arxiv.org/abs/2209.14734

Abstract: This work introduces DiGress, a discrete denoising diffusion model for generating graphs with categorical node and edge attributes. Our model defines a diffusion process that progressively edits a graph with noise (adding or removing edges, changing the categories), and a graph transformer network that learns to revert this process. With these two ingredients in place, we reduce distribution learning over graphs to a simple sequence of classification tasks. We further improve sample quality by proposing a new Markovian noise model that preserves
the marginal distribution of node and edge types during diffusion, and by adding auxiliary graph-theoretic features derived from the noisy graph at each diffusion step. Finally, we propose a guidance procedure for conditioning the generation on graph-level features. Overall, DiGress achieves state-of-the-art performance on both molecular and non-molecular datasets, with up to 3x validity improvement on a dataset of planar graphs. In particular, it is the first model that scales to the large GuacaMol dataset containing 1.3M drug-like molecules without using a molecule-specific representation such as SMILES or fragments.

Authors: Clement Vignac, Igor Krawczuk, Antoine Siraudin, Bohan Wang, Volkan Cevher, Pascal Frossard

Twitter Hannes: Twitter: HannesStaerk
Twitter Dominique: Twitter: dom_beaini
Twitter Valence Discovery: Twitter: valence_ai

Reading Group Slack: https://join.slack.com/t/logag/shared...

~

Chapters

00:00 - Intro
01:27 - Denoising Diffusion Models
05:47 - The Price of Efficiency
07:03 - Discrete Diffusion
10:13 - DiGress: Discrete Graph Denoting Diffusion
13:03 - DiGress: Challenges and Training Methods
30:11 - DiGress: Denoising Network
32:01 -  DiGress: Properties
33:31 - Results
34:44 - Improving DiGress with Marginal Transitions and Structural Features
41:15 - Final Results
49:14 - Summary and What’s Next
56:09 - Q+A