Junction tree variational autoencoder for molecular graph generation
Name
1802.04364.pdf
Description
Accepted version
Size
2.28 MB
Format
Adobe PDF
Checksum (MD5)
c85c5bcf21ca347fdabf742557cdedbc
Author(s)
Jin, Wengong
Barzilay, Regina
Jaakkola, Tommi
Date Issued
2018
Citation
Jin, Wengong, Barzilay, Regina and Jaakkola, Tommi. 2018. "Junction tree variational autoencoder for molecular graph generation."
Version
Author's final manuscript
Abstract
© 2018 by authors.All right reserved. We seek to automate the design of molecules based on specific chemical properties. In computational terms, this task involves continuous embedding and generation of molecular graphs. Our primary contribution is the direct realization of molecular graphs, a task previously approached by generating linear SMILES strings instead of graphs. Our junction tree variational autoencoder generates molecular graphs in two phases, by first generating a tree-structured scaffold over chemical substructures, and then combining them into a molecule with a graph message passing network. This approach allows us to incrementally expand molecules while maintaining chemical validity at every step. We evaluate our model on multiple tasks ranging from molecular generation to optimization. Across these tasks, our model outperforms previous state-of-the-art baselines by a significant margin.
MIT Department
Massachusetts Institute of Technology. Computer Science and Artificial Intelligence Laboratory
Terms of Use
Creative Commons Attribution-Noncommercial-Share Alike
Persistent DSpace Link
