Deep neural networks (‘deep learning’) have emerged as a technology of choice to tackle problems in speech recognition, computer vision, finance, etc. However, adoption of deep learning in physical domains brings substantial challenges stemming from the correlative nature of deep learning methods compared to the causal, hypothesis driven nature of modern science. We argue that the broad adoption of Bayesian methods incorporating prior knowledge, development of solutions with incorporated physical constraints and parsimonious structural descriptors and generative models, and ultimately adoption of causal models, offers a path forward for fundamental and applied research.
Vasudevan, Rama K., et al. "Off-the-shelf deep learning is not enough, and requires parsimony, Bayesianity, and causality." npj Computational Materials, vol. 7, no. 1, Jan. 2021. https://doi.org/10.1038/s41524-020-00487-0
Vasudevan, Rama K., Ziatdinov, Maxim, Vlcek, Lukas, & Kalinin, Sergei V. (2021). Off-the-shelf deep learning is not enough, and requires parsimony, Bayesianity, and causality. npj Computational Materials, 7(1). https://doi.org/10.1038/s41524-020-00487-0
Vasudevan, Rama K., Ziatdinov, Maxim, Vlcek, Lukas, et al., "Off-the-shelf deep learning is not enough, and requires parsimony, Bayesianity, and causality," npj Computational Materials 7, no. 1 (2021), https://doi.org/10.1038/s41524-020-00487-0
@article{osti_1763073,
author = {Vasudevan, Rama K. and Ziatdinov, Maxim and Vlcek, Lukas and Kalinin, Sergei V.},
title = {Off-the-shelf deep learning is not enough, and requires parsimony, Bayesianity, and causality},
annote = {Abstract Deep neural networks (‘deep learning’) have emerged as a technology of choice to tackle problems in speech recognition, computer vision, finance, etc. However, adoption of deep learning in physical domains brings substantial challenges stemming from the correlative nature of deep learning methods compared to the causal, hypothesis driven nature of modern science. We argue that the broad adoption of Bayesian methods incorporating prior knowledge, development of solutions with incorporated physical constraints and parsimonious structural descriptors and generative models, and ultimately adoption of causal models, offers a path forward for fundamental and applied research.},
doi = {10.1038/s41524-020-00487-0},
url = {https://www.osti.gov/biblio/1763073},
journal = {npj Computational Materials},
issn = {ISSN 2057-3960},
number = {1},
volume = {7},
place = {United Kingdom},
publisher = {Nature Publishing Group},
year = {2021},
month = {01}}
Medical Image Computing and Computer-Assisted Intervention – MICCAI 2015: 18th International Conference, Munich, Germany, October 5-9, 2015, Proceedings, Part IIIhttps://doi.org/10.1007/978-3-319-24574-4_28