{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,6,29]],"date-time":"2026-06-29T21:33:26Z","timestamp":1782768806345,"version":"3.54.5"},"reference-count":31,"publisher":"IOP Publishing","issue":"3","license":[{"start":{"date-parts":[[2021,7,13]],"date-time":"2021-07-13T00:00:00Z","timestamp":1626134400000},"content-version":"vor","delay-in-days":0,"URL":"http:\/\/creativecommons.org\/licenses\/by\/4.0"},{"start":{"date-parts":[[2021,7,13]],"date-time":"2021-07-13T00:00:00Z","timestamp":1626134400000},"content-version":"tdm","delay-in-days":0,"URL":"https:\/\/iopscience.iop.org\/info\/page\/text-and-data-mining"}],"funder":[{"name":"Canada 150 Research Chair Program"},{"name":"Anders G Froseth"},{"DOI":"10.13039\/501100007220","name":"Tata Steel","doi-asserted-by":"crossref","id":[{"id":"10.13039\/501100007220","id-type":"DOI","asserted-by":"crossref"}]},{"name":"Austrian Science Fund"},{"DOI":"10.13039\/100000006","name":"Office of Naval Research","doi-asserted-by":"crossref","id":[{"id":"10.13039\/100000006","id-type":"DOI","asserted-by":"crossref"}]}],"content-domain":{"domain":["iopscience.iop.org"],"crossmark-restriction":false},"short-container-title":["Mach. Learn.: Sci. Technol."],"published-print":{"date-parts":[[2021,9,1]]},"abstract":"<jats:title>Abstract<\/jats:title>\n               <jats:p>Computer-based <jats:italic>de-novo<\/jats:italic> design of functional molecules is one of the most prominent challenges in cheminformatics today. As a result, generative and evolutionary inverse designs from the field of artificial intelligence have emerged at a rapid pace, with aims to optimize molecules for a particular chemical property. These models \u2018indirectly\u2019 explore the chemical space; by learning latent spaces, policies, and distributions, or by applying mutations on populations of molecules. However, the recent development of the SELFIES (Krenn 2020 <jats:italic>Mach. Learn.: Sci. Technol.<\/jats:italic> \n                  <jats:bold>1<\/jats:bold> 045024) string representation of molecules, a surjective alternative to SMILES, have made possible other potential techniques. Based on SELFIES, we therefore propose PASITHEA, a direct gradient-based molecule optimization that applies inceptionism (Mordvintsev 2015) techniques from computer vision. PASITHEA exploits the use of gradients by directly reversing the learning process of a neural network, which is trained to predict real-valued chemical properties. Effectively, this forms an inverse regression model, which is capable of generating molecular variants optimized for a certain property. Although our results are preliminary, we observe a shift in distribution of a chosen property during inverse-training, a clear indication of PASITHEA\u2019s viability. A striking property of inceptionism is that we can directly probe the model\u2019s <jats:italic>understanding<\/jats:italic> of the chemical space on which it is trained. We expect that extending PASITHEA to larger datasets, molecules and more complex properties will lead to advances in the design of new functional molecules as well as the interpretation and explanation of machine learning models.<\/jats:p>","DOI":"10.1088\/2632-2153\/ac09d6","type":"journal-article","created":{"date-parts":[[2021,6,9]],"date-time":"2021-06-09T22:45:35Z","timestamp":1623278735000},"page":"03LT02","update-policy":"https:\/\/doi.org\/10.1088\/crossmark-policy","source":"Crossref","is-referenced-by-count":36,"title":["Deep molecular dreaming: inverse machine learning for de-novo molecular design and interpretability with surjective representations"],"prefix":"10.1088","volume":"2","author":[{"ORCID":"https:\/\/orcid.org\/0000-0003-3266-2634","authenticated-orcid":false,"given":"Cynthia","family":"Shen","sequence":"first","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0003-1620-9207","authenticated-orcid":false,"given":"Mario","family":"Krenn","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Sagi","family":"Eppel","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-8277-4434","authenticated-orcid":false,"given":"Al\u00e1n","family":"Aspuru-Guzik","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"266","published-online":{"date-parts":[[2021,7,13]]},"reference":[{"key":"mlstac09d6bib1","article-title":"Very deep convolutional networks for large-scale image recognition","author":"Simonyan","year":"2014"},{"key":"mlstac09d6bib2","author":"Linder-Noren","year":"2019"},{"key":"mlstac09d6bib3","doi-asserted-by":"publisher","first-page":"360","DOI":"10.1126\/science.aat2663","article-title":"Inverse molecular design using machine learning: generative models for matter engineering","volume":"361","author":"Sanchez-Lengeling","year":"2018","journal-title":"Science"},{"key":"mlstac09d6bib4","doi-asserted-by":"publisher","first-page":"133","DOI":"10.1016\/j.trechm.2020.11.004","article-title":"Defining and exploring chemical spaces","volume":"3","author":"Coley","year":"2021","journal-title":"Trends in Chemistry"},{"key":"mlstac09d6bib5","doi-asserted-by":"publisher","first-page":"268","DOI":"10.1021\/acscentsci.7b00572","article-title":"Automatic chemical design using a data-driven continuous representation of molecules","volume":"4","author":"G\u00f3mez-Bombarelli","year":"2018","journal-title":"ACS Central Sci."},{"key":"mlstac09d6bib6","article-title":"Junction tree variational autoencoder for molecular graph generation","author":"Jin","year":"2018"},{"key":"mlstac09d6bib7","article-title":"Constrained generation of semantically valid graphs via regularizing variational autoencoders","author":"Tengfei","year":"2018"},{"key":"mlstac09d6bib8","article-title":"Objective-reinforced generative adversarial networks (organ) for sequence generation models","author":"Guimaraes","year":"2017"},{"key":"mlstac09d6bib9","article-title":"Molgan: an implicit generative model for small molecular graphs","author":"Nicola","year":"2018"},{"key":"mlstac09d6bib10","doi-asserted-by":"publisher","first-page":"1","DOI":"10.1038\/s41598-019-47148-x","article-title":"Optimization of molecules via deep reinforcement learning","volume":"9","author":"Zhou","year":"2019","journal-title":"Sci. Rep."},{"key":"mlstac09d6bib11","first-page":"pp 6410","author":"You","year":"2018"},{"key":"mlstac09d6bib12","doi-asserted-by":"publisher","first-page":"3567","DOI":"10.1039\/C8SC05372C","article-title":"A graph-based genetic algorithm and generative model\/Monte Carlo tree search for the exploration of chemical space","volume":"10","author":"Jensen","year":"2019","journal-title":"Chem. Sci."},{"key":"mlstac09d6bib13","article-title":"Augmenting genetic algorithms with deep neural networks for exploring the chemical space","author":"Nigam","year":"2019"},{"key":"mlstac09d6bib14","doi-asserted-by":"publisher","first-page":"e11","DOI":"10.7717\/peerj-pchem.11","article-title":"Chemical space exploration: how genetic algorithms find the needle in the haystack","volume":"2","author":"Henault","year":"2020","journal-title":"PeerJ. Phys. Chem."},{"key":"mlstac09d6bib15","doi-asserted-by":"publisher","first-page":"e17","DOI":"10.1002\/ail2.17","article-title":"Assessing methods and obstacles in chemical space exploration","volume":"1","author":"Reeves","year":"2020","journal-title":"Appl. AI Lett."},{"key":"mlstac09d6bib16","article-title":"Inceptionism: going deeper into neural networks","author":"Mordvintsev","year":"2015"},{"key":"mlstac09d6bib17","doi-asserted-by":"publisher","DOI":"10.1088\/2632-2153\/aba947","article-title":"Self-referencing embedded strings (selfies): a 100% robust molecular string representation","volume":"1","author":"Krenn","year":"2020","journal-title":"Mach. Learn.: Sci. Technol."},{"key":"mlstac09d6bib18","author":"Landrum","year":"2006"},{"key":"mlstac09d6bib19","doi-asserted-by":"publisher","first-page":"337","DOI":"10.1016\/j.ddtec.2004.11.007","article-title":"Lead-and drug-like compounds: the rule-of-five revolution","volume":"1","author":"Lipinski","year":"2004","journal-title":"Drug Discovery Today: Technol."},{"key":"mlstac09d6bib20","article-title":"Beyond generative models: superfast traversal, optimization, novelty, exploration and discovery (stoned) algorithm for molecules using selfies","author":"Nigam","year":"2020","journal-title":"ChemRxiv"},{"key":"mlstac09d6bib21","doi-asserted-by":"publisher","first-page":"52138","DOI":"10.1109\/ACCESS.2018.2870052","article-title":"Peeking inside the black-box: a survey on explainable artificial intelligence (XAI)","volume":"6","author":"Adadi","year":"2018","journal-title":"IEEE Access"},{"key":"mlstac09d6bib22","doi-asserted-by":"publisher","DOI":"10.1103\/PhysRevLett.124.010508","article-title":"Discovering physical concepts with neural networks","volume":"124","author":"Iten","year":"2020","journal-title":"Phys. Rev. Lett."},{"key":"mlstac09d6bib23","doi-asserted-by":"publisher","first-page":"42200","DOI":"10.1109\/ACCESS.2020.2976199","article-title":"Explainable machine learning for scientific insights and discoveries","volume":"8","author":"Roscher","year":"2020","journal-title":"IEEE Access"},{"key":"mlstac09d6bib24","article-title":"Scientific intuition inspired by machine learning generated hypotheses","author":"Friederich","year":"2020"},{"key":"mlstac09d6bib25","doi-asserted-by":"publisher","first-page":"7526","DOI":"10.1364\/OE.419138","article-title":"Neural-adjoint method for the inverse design of all-dielectric metasurfaces","volume":"29","author":"Deng","year":"2021","journal-title":"Opt. Express"},{"key":"mlstac09d6bib26","article-title":"Benchmarking deep inverse models over time, and the neural-adjoint method","author":"Ren","year":"2020"},{"key":"mlstac09d6bib27","doi-asserted-by":"publisher","first-page":"105","DOI":"10.1038\/s41567-020-1018-2","article-title":"Machine learning the thermodynamic arrow of time","volume":"17","author":"Seif","year":"2021","journal-title":"Nat. Phys."},{"key":"mlstac09d6bib28","first-page":"pp 5188","article-title":"Understanding deep image representations by inverting them","author":"Mahendran","year":"2015"},{"key":"mlstac09d6bib29","volume":"20.323l)","author":"Stewart","year":"2020"},{"key":"mlstac09d6bib30","doi-asserted-by":"publisher","first-page":"1","DOI":"10.1007\/s00894-012-1667-x","article-title":"Optimization of parameters for semiempirical methods VI: more modifications to the NDDO approximations and re-optimization of parameters","volume":"19","author":"Stewart","year":"2013","journal-title":"J. Mol. Model."},{"key":"mlstac09d6bib31","author":"","year":"2019"}],"container-title":["Machine Learning: Science and Technology"],"original-title":[],"link":[{"URL":"https:\/\/iopscience.iop.org\/article\/10.1088\/2632-2153\/ac09d6","content-type":"text\/html","content-version":"am","intended-application":"text-mining"},{"URL":"https:\/\/iopscience.iop.org\/article\/10.1088\/2632-2153\/ac09d6\/pdf","content-type":"application\/pdf","content-version":"am","intended-application":"text-mining"},{"URL":"https:\/\/iopscience.iop.org\/article\/10.1088\/2632-2153\/ac09d6","content-type":"text\/html","content-version":"vor","intended-application":"text-mining"},{"URL":"https:\/\/iopscience.iop.org\/article\/10.1088\/2632-2153\/ac09d6\/pdf","content-type":"application\/pdf","content-version":"vor","intended-application":"text-mining"},{"URL":"https:\/\/iopscience.iop.org\/article\/10.1088\/2632-2153\/ac09d6\/pdf","content-type":"application\/pdf","content-version":"am","intended-application":"syndication"},{"URL":"https:\/\/iopscience.iop.org\/article\/10.1088\/2632-2153\/ac09d6\/pdf","content-type":"application\/pdf","content-version":"vor","intended-application":"syndication"},{"URL":"https:\/\/iopscience.iop.org\/article\/10.1088\/2632-2153\/ac09d6\/pdf","content-type":"application\/pdf","content-version":"am","intended-application":"similarity-checking"},{"URL":"https:\/\/iopscience.iop.org\/article\/10.1088\/2632-2153\/ac09d6\/pdf","content-type":"application\/pdf","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2021,12,13]],"date-time":"2021-12-13T16:06:27Z","timestamp":1639411587000},"score":1,"resource":{"primary":{"URL":"https:\/\/iopscience.iop.org\/article\/10.1088\/2632-2153\/ac09d6"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2021,7,13]]},"references-count":31,"journal-issue":{"issue":"3","published-online":{"date-parts":[[2021,7,13]]},"published-print":{"date-parts":[[2021,9,1]]}},"URL":"https:\/\/doi.org\/10.1088\/2632-2153\/ac09d6","relation":{},"ISSN":["2632-2153"],"issn-type":[{"value":"2632-2153","type":"electronic"}],"subject":[],"published":{"date-parts":[[2021,7,13]]},"assertion":[{"value":"Deep molecular dreaming: inverse machine learning for de-novo molecular design and interpretability with surjective representations","name":"article_title","label":"Article Title"},{"value":"Machine Learning: Science and Technology","name":"journal_title","label":"Journal Title"},{"value":"paper","name":"article_type","label":"Article Type"},{"value":"\u00a9 2021 The Author(s). Published by IOP Publishing Ltd","name":"copyright_information","label":"Copyright Information"},{"value":"2021-02-20","name":"date_received","label":"Date Received","group":{"name":"publication_dates","label":"Publication dates"}},{"value":"2021-06-09","name":"date_accepted","label":"Date Accepted","group":{"name":"publication_dates","label":"Publication dates"}},{"value":"2021-07-13","name":"date_epub","label":"Online publication date","group":{"name":"publication_dates","label":"Publication dates"}}]}}