% IMPORTANT: The following is UTF-8 encoded. This means that in the presence
% of non-ASCII characters, it will not work with BibTeX 0.99 or older.
% Instead, you should use an up-to-date BibTeX implementation like “bibtex8” or
% “biber”.
@ARTICLE{Schweidtmann:917552,
author = {Schweidtmann, Artur M. and Rittig, Jan G. and Weber, Jana
M. and Grohe, Martin and Dahmen, Manuel and Leonhard, Kai
and Mitsos, Alexander},
title = {{P}hysical {P}ooling {F}unctions in {G}raph {N}eural
{N}etworks for {M}olecular {P}roperty {P}rediction},
publisher = {arXiv},
reportid = {FZJ-2023-00754},
year = {2022},
abstract = {Graph neural networks (GNNs) are emerging in chemical
engineering for the end-to-end learning of physicochemical
properties based on molecular graphs. A key element of GNNs
is the pooling function which combines atom feature vectors
into molecular fingerprints. Most previous works use a
standard pooling function to predict a variety of
properties. However, unsuitable pooling functions can lead
to unphysical GNNs that poorly generalize. We compare and
select meaningful GNN pooling methods based on physical
knowledge about the learned properties. The impact of
physical pooling functions is demonstrated with molecular
properties calculated from quantum mechanical computations.
We also compare our results to the recent set2set pooling
approach. We recommend using sum pooling for the prediction
of properties that depend on molecular size and compare
pooling functions for properties that are molecular
size-independent. Overall, we show that the use of physical
pooling functions significantly enhances generalization.},
keywords = {Machine Learning (cs.LG) (Other) / FOS: Computer and
information sciences (Other)},
cin = {IEK-10},
cid = {I:(DE-Juel1)IEK-10-20170217},
pnm = {1121 - Digitalization and Systems Technology for
Flexibility Solutions (POF4-112)},
pid = {G:(DE-HGF)POF4-1121},
typ = {PUB:(DE-HGF)25},
doi = {10.48550/ARXIV.2207.13779},
url = {https://juser.fz-juelich.de/record/917552},
}