We gratefully acknowledge support from
the Simons Foundation and member institutions.
Full-text links:


Current browse context:


Change to browse by:

References & Citations


(what is this?)
CiteULike logo BibSonomy logo Mendeley logo del.icio.us logo Digg logo Reddit logo ScienceWISE logo

Quantitative Biology > Biomolecules

Title: A Systematic Comparison Study on Hyperparameter Optimisation of Graph Neural Networks for Molecular Property Prediction

Abstract: Graph neural networks (GNNs) have been proposed for a wide range of graph-related learning tasks. In particular, in recent years, an increasing number of GNN systems were applied to predict molecular properties. However, a direct impediment is to select appropriate hyperparameters to achieve satisfactory performance with lower computational cost. Meanwhile, many molecular datasets are far smaller than many other datasets in typical deep learning applications. Most hyperparameter optimization (HPO) methods have not been explored in terms of their efficiencies on such small datasets in the molecular domain. In this paper, we conducted a theoretical analysis of common and specific features for two state-of-the-art and popular algorithms for HPO: TPE and CMA-ES, and we compared them with random search (RS), which is used as a baseline. Experimental studies are carried out on several benchmarks in MoleculeNet, from different perspectives to investigate the impact of RS, TPE, and CMA-ES on HPO of GNNs for molecular property prediction. In our experiments, we concluded that RS, TPE, and CMA-ES have their individual advantages in tackling different specific molecular problems. Finally, we believe our work will motivate further research on GNN as applied to molecular machine learning problems in chemistry and materials sciences.
Subjects: Biomolecules (q-bio.BM); Machine Learning (cs.LG)
DOI: 10.1145/3449639.3459370
Cite as: arXiv:2102.04283 [q-bio.BM]
  (or arXiv:2102.04283v3 [q-bio.BM] for this version)

Submission history

From: Yingfang Yuan [view email]
[v1] Mon, 8 Feb 2021 15:40:50 GMT (1582kb)
[v2] Thu, 15 Apr 2021 22:33:33 GMT (1025kb)
[v3] Thu, 22 Apr 2021 00:46:07 GMT (929kb)

Link back to: arXiv, form interface, contact.