DiscoverMaterials and MegabytesPaper Interview - Learning from Failure: Predicting Electronic Structure Calculation Outcomes with Machine Learning Models
Paper Interview - Learning from Failure: Predicting Electronic Structure Calculation Outcomes with Machine Learning Models

Paper Interview - Learning from Failure: Predicting Electronic Structure Calculation Outcomes with Machine Learning Models

Update: 2020-01-13
Share

Description

We discuss the paper Learning from Failure: Predicting Electronic Structure Calculation Outcomes with Machine Learning Models with the author Prof. Heather J. Kulik.

Papers discussed in this episode:

  • (Main discussion) Duan, C.; Janet, J. P.; Liu, F.; Nandy, A.; Kulik, H. J. Learning from Failure: Predicting Electronic Structure Calculation Outcomes with Machine Learning Models. J. Chem. Theory Comput. 2019, 15 (4), 2331–2345. https://doi.org/10.1021/acs.jctc.9b00057.
  • (More on uncertainty metrics in latent space) Janet, J. P.; Duan, C.; Yang, T.; Nandy, A.; Kulik, H. J. A Quantitative Uncertainty Metric Controls Error in Neural Network-Driven Chemical Discovery. Chem. Sci. 2019, 10 (34), 7913–7922. https://doi.org/10.1039/C9SC02298H.
  • (Follow-up paper with active learning) Janet, J. P.; Ramesh, S.; Duan, C.; Kulik, H. Accurate Multi-Objective Design in a Space of Millions of Transition Metal Complexes with Neural-Network-Driven Efficient Global Optimization. 2019. https://doi.org/10.26434/chemrxiv.11367572.v1.

Kulik group website: http://hjkgrp.mit.edu/

Comments 
00:00
00:00
x

0.5x

0.8x

1.0x

1.25x

1.5x

2.0x

3.0x

Sleep Timer

Off

End of Episode

5 Minutes

10 Minutes

15 Minutes

30 Minutes

45 Minutes

60 Minutes

120 Minutes

Paper Interview - Learning from Failure: Predicting Electronic Structure Calculation Outcomes with Machine Learning Models

Paper Interview - Learning from Failure: Predicting Electronic Structure Calculation Outcomes with Machine Learning Models

Stanford Materials Computation and Theory Group, Qian Yang's lab at the University of Connecticut