Bond type restricted radial distribution functions for accurate machine\n learning prediction of atomization energies

2018/07/26 by Mykhaylo Krykunov, Krykunov, Mykhaylo, Tom K. Woo +1 · 1 citation
Biochemistry, Genetics and Molecular Biology · Computer Science · Materials Science · Physics and Astronomy · #Advanced Chemical Physics Studies #Chemical Physics (physics.chem-ph) #Computational Drug Discovery Methods #FOS: Physical sciences #Machine Learning in Materials Science #Protein Structure and Dynamics

paper · pdf · doi:10.48550/arxiv.1807.10301

openalex publication_date 2018/07/26 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28

Abstract

Understanding the performance of machine learning algorithms is essential for\ndesigning more accurate and efficient statistical models. It is not always\npossible to unravel the reasoning of neural networks. Here we propose a method\nfor calculating machine learning kernels in closed and analytic form by\ncombining atomic property weighted radial distribution function (AP-RDF)\ndescriptor with a Gaussian kernel. This allowed us to analyse and improve the\nperformance of the Bag-of-Bonds descriptor, when the bond type restriction is\nincluded in AP-RDF. The improvement is achieved for the prediction of molecular\natomization energies and is due to the incorporation of a tensor product into\nthe kernel which captures the multidimensional representation of the AP-RDF. On\nthe other hand, the numerical version of the AP-RDF is a constant size\ndescriptor, and it is more computationally efficient than Bag-of-Bonds. We have\nalso discussed a connection between molecular quantum similarity and machine\nlearning kernels with first principles kind of descriptors.\n

Citations

• Die Berechnung optischer und elektrostatischer Gitterpotentiale
• Fast and Accurate Modeling of Molecular Atomization Energies with Machine Learning
• Crystal structure representations for machine learning models of formation energies

Cited by

• Physical regularized Hierarchical Generative Model for Metallic Glass Structural Generation and Energy Prediction

Related

• Fourier series of atomic radial distribution functions: A molecular fingerprint for machine learning models of quantum chemical properties
• Stacked Generalization Approach to Improve Prediction of Molecular Atomization Energies
• Atom-Density Representations for Machine Learning
• Physical machine learning outperforms "human learning" in Quantum Chemistry
• On the Interplay of Subset Selection and Informed Graph Neural Networks