2018/12/11 by Kristof T. Schütt, Alexandre Tkatchenko, Schütt, Kristof T. +4 · 2 citations
Chemistry · Computer Science · Materials Science · Mathematics · Physics and Astronomy · Psychology · #Artificial intelligence #Artificial neural network #Chemical space #Chemistry #Cognitive science #Computational Drug Discovery Methods #Computational Physics (physics.comp-ph) #Computer science #Convolutional neural network #FOS: Computer and information sciences #FOS: Physical sciences #Intuition #Machine Learning (cs.LG) #Machine Learning (stat.ML) #Machine Learning in Materials Science #Psychology #Space (punctuation) #Theoretical computer science #Variety (cybernetics) #Various Chemistry Research Topics #cs.LG #physics.comp-ph #stat.ML
paper · pdf · doi:10.48550/arxiv.1812.04690
published in arXiv (Cornell University) (Cornell University)
arxiv created 2018/12/11 · openalex publication_date 2018/12/11 · arxiv updated 2018/12/13 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28
Deep Learning has been shown to learn efficient representations for\nstructured data such as image, text or audio. In this chapter, we present\nneural network architectures that are able to learn efficient representations\nof molecules and materials. In particular, the continuous-filter convolutional\nnetwork SchNet accurately predicts chemical properties across compositional and\nconfigurational space on a variety of datasets. Beyond that, we analyze the\nobtained representations to find evidence that their spatial and chemical\nproperties agree with chemical intuition.\n