Learning Invariant Representations of Molecules for Atomization Energy Prediction

By Grgoire Montavon, Katja Hansen, Siamac Fazli, Matthias Rupp , Franziska Biegler, Andreas Ziehe, Alexandre Tkatchenko, A. V. Lilienfeld, K. Müller. 2012

136 citations recorded by the source corpus.

Subfields: Representation & Self-Supervised Learning, Computational Chemistry & Drug Discovery

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Representation & Self-Supervised Learning

In Computational Chemistry & Drug Discovery

Citations in the corpus

Neural Message Passing for Quantum Chemistry (2017, background): “Correspondence to: Justin Gilmer”
Spherical CNNs (2018, methodology): “To this end Rupp et al. (2012) propose a distance measure between Coulomb matrices used within Gaussian kernels whereas Montavon et al. (2012) propose sorting C or random sampling index permutations.”
Scalable Kernel Methods via Doubly Stochastic Gradients (2014, methodology): “We test our algorithm for kernel ridge regression with neural network proposed in [25] on two large-scale real-world regression datasets, (8) and (9) in Table 3.”
Neural Networks for the Prediction of Organic Chemistry Reactions (2016, unknown): “29 This last example is particularly interesting as Go is a complex board game with a search space of 10 170 , which is on the order of chemical space for medium sized molecules.”
Molecular Property Prediction: A Multilevel Quantum Interactions Modeling Perspective (2019, methodology): “Generally, they rely on rulebased hand crafted features using the domain knowledge of physics and chemistry, including bag of bonds, coulomb matrix, and histogram of distances, angles and dihedral angles (Huang and von Lilienfeld 2016; Hansen et al. 2015; Montavon et al. 2012).”
Learning from Conditional Distributions via Dual Kernel Embeddings (2016, methodology): “ we test the algorithm on two tasks. We rst apply the algorithm to robust learning problem where the inputs are contaminated, and then, we conduct comparison on molecular eneretics prediction problem [19]. We compare the proposed algorithm with SGD with virtual samples technique [24, 18] and SGD wit”
Hunt For The Unique, Stable, Sparse And Fast Feature Learning On Graphs (2017, background): “A tangential work [24] related to constructing features based on atoms 3D space coordinates rather than operating on a graph structure, can be also considered in this category.”
Learning from Conditional Distributions via Dual Embeddings (2016, methodology): “We first apply the algorithm to robust learning problem where the inputs are contaminated, and then, we conduct comparison on a molecular energetics prediction problem (Montavon et al., 2012).”
DScribe: Library of Descriptors for Machine Learning in Materials Science (2019, background): “ncode the matrices by their eigenvalues, which are invariant to changes in the column and row ordering [9]. Another way is to order the rows and columns by a chosen norm, typically the Euclidean norm [33]. A third approach is to augment the dataset by creating multiple slightly varying matrices for ”
Modeling the formation and growth of atmospheric molecular clusters: A review (2020, background): “Geometry representing descriptors include Coulomb Matrix (Montavon et al., 2015; Rupp et al., 2012) , Bag of Bonds (Hansen et al., 2015), Many-Body Tensor Representation (Huo & Rupp, 2017), Atom-Centered Symmetry Functions (Behler, 2011), or Smooth Overlap of Atomic Positions (Bartók et al., 2013;…”
Graph Capsule Convolutional Neural Networks (2018, background): “Another line of research in this area focuses on efﬁciently computing these kernels either through exploiting certain structure dependency, or via approximation or randomization (Feragen et al., 2013; de Vries, 2013; Neumann et al., 2012).”
Quantum chemical accuracy from density functional approximations via machine learning (2020, background): “(2) based on a set of 102 random water geometries (Fig.”