Analysis of Atomistic Representations Using Weighted Skip-Connections
Kim A. Nicoli, Pan Kessel, Michael Gastegger, Kristof T. Schütt
2018-10-23BIG-bench Machine Learning
Abstract
In this work, we extend the SchNet architecture by using weighted skip connections to assemble the final representation. This enables us to study the relative importance of each interaction block for property prediction. We demonstrate on both the QM9 and MD17 dataset that their relative weighting depends strongly on the chemical composition and configurational degrees of freedom of the molecules which opens the path towards a more detailed understanding of machine learning models for molecules.
Related Papers
(1,1)-Cluster Editing is Polynomial-time Solvable2022-10-14AnalogVNN: A fully modular framework for modeling and optimizing photonic neural networks2022-10-14Discover the Mysteries of the Maya: Selected Contributions from the Machine Learning Challenge & The Discovery Challenge Workshop at ECML PKDD 20212022-08-05Tailoring to the Tails: Risk Measures for Fine-Grained Tail Sensitivity2022-08-05Explanation of Machine Learning Models of Colon Cancer Using SHAP Considering Interaction Effects2022-08-05Machine Learning and Bioinformatics for Diagnosis Analysis of Obesity Spectrum Disorders2022-08-05A Lightweight Machine Learning Pipeline for LiDAR-simulation2022-08-05Evolutionary bagging for ensemble learning2022-08-04