A Caltech Library Service

Deep Reasoning Networks for Unsupervised Pattern De-mixing with Constraint Reasoning

Chen, Di and Bai, Yiwei and Zhao, Wenting and Ament, Sebastian and Gregoire, John M. and Gomes, Carla P. (2020) Deep Reasoning Networks for Unsupervised Pattern De-mixing with Constraint Reasoning. Proceedings of Machine Learning Research, 119 . pp. 1500-1509. ISSN 2640-3498.

[img] PDF - Published Version
See Usage Policy.


Use this Persistent URL to link to this item:


We introduce Deep Reasoning Networks (DRNets), an end-to-end framework that combines deep learning with constraint reasoning for solving pattern de-mixing problems, typically in an unsupervised or very-weakly-supervised setting. DRNets exploit problem structure and prior knowledge by tightly combining constraint reasoning with stochastic-gradient-based neural network optimization. Our motivating task is from materials discovery and concerns inferring crystal structures of materials from X-ray diffraction data (Crystal-Structure-Phase-Mapping). Given the complexity of its underlying scientific domain, we start by introducing DRNets on an analogous but much simpler task: de-mixing overlapping hand-written Sudokus (Multi-MNIST-Sudoku). On Multi-MNIST-Sudoku, DRNets almost perfectly recovered the mixed Sudokus’ digits, with 100% digit accuracy, outperforming the supervised state-of-the-art MNIST de-mixing models. On Crystal-Structure-Phase-Mapping, DRNets significantly outperform the state of the art and experts’ capabilities, recovering more precise and physically meaningful crystal structures.

Item Type:Article
Related URLs:
URLURL TypeDescription
Ament, Sebastian0000-0001-6316-4633
Gregoire, John M.0000-0002-2863-5265
Gomes, Carla P.0000-0002-4441-7225
Additional Information:© 2020 by the author(s). This research was supported by NSF awards CCF-1522054 (Expeditions in computing) and CNS-1059284 (Infrastructure), AFOSR Multidisciplinary University Research Initiatives (MURI) Program FA9550-18-1-0136, ARO awards W911NF-14-1-0498 and W911NF-17-1-0187, US DOE Award No. DE-SC0020383, and an award from the Toyota Research Institute. Materials science experiments were supported by US DOE Award No. DE-SC0004993. Use of SSRL is supported by DOE Contract No. DE-AC02-76SF00515. We are grateful for the assistance of Junwen Bai for running the IAFD baseline and Aniketa Shinde for photoelectrochemistry experiments.
Funding AgencyGrant Number
Air Force Office of Scientific Research (AFOSR)FA9550-18-1-0136
Army Research Office (ARO)W911NF-14-1-0498
Toyota Research InstituteUNSPECIFIED
Department of Energy (DOE)DE-SC0004993
Department of Energy (DOE)DE-AC02-76SF00515
Record Number:CaltechAUTHORS:20211008-163254045
Persistent URL:
Official Citation:Chen, D., Bai, Y., Zhao, W., Ament, S., Gregoire, J., Gomes, C. (2020). Deep Reasoning Networks for Unsupervised Pattern De-mixing with Constraint Reasoning. Proceedings of the 37th International Conference on Machine Learning, in Proceedings of Machine Learning Research, 119:1500-1509; Available from
Usage Policy:No commercial reproduction, distribution, display or performance rights in this work are provided.
ID Code:111279
Deposited By: Tony Diaz
Deposited On:08 Oct 2021 19:10
Last Modified:08 Oct 2021 19:10

Repository Staff Only: item control page