A reproducibility study of "Augmenting Genetic Algorithms with Deep Neural Networks for Exploring the Chemical Space"

Nigam et al. reported a genetic algorithm (GA) utilizing the SELFIES representation and also propose an adaptive, neural network-based penalty that is supposed to improve the diversity of the generated molecules. The main claims of the paper are that this GA outperforms other generative techniques (...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Jablonka, Kevin Maik, Mcilwaine, Fergus, Garcia, Susana, Smit, Berend, Yoo, Brian
Format: Artikel
Sprache:eng
Schlagworte:
Online-Zugang:Volltext bestellen
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
container_end_page
container_issue
container_start_page
container_title
container_volume
creator Jablonka, Kevin Maik
Mcilwaine, Fergus
Garcia, Susana
Smit, Berend
Yoo, Brian
description Nigam et al. reported a genetic algorithm (GA) utilizing the SELFIES representation and also propose an adaptive, neural network-based penalty that is supposed to improve the diversity of the generated molecules. The main claims of the paper are that this GA outperforms other generative techniques (as measured by the penalized logP) and that a neural network-based adaptive penalty increases the diversity of the generated molecules. In this work, we investigated the reproducibility of their claims. Overall, we were able to reproduce comparable results using the SELFIES-based GA, but mostly by exploiting deficiencies of the (easily optimizable) fitness function (i.e., generating long, sulfur containing chains). In addition, we reproduce results showing that the discriminator can be used to bias the generation of molecules to ones that are similar to the reference set. Lastly, we attempted to quantify the evolution of the diversity, understand the influence of some hyperparameters, and propose improvements to the adaptive penalty.
doi_str_mv 10.48550/arxiv.2102.00700
format Article
fullrecord <record><control><sourceid>arxiv_GOX</sourceid><recordid>TN_cdi_arxiv_primary_2102_00700</recordid><sourceformat>XML</sourceformat><sourcesystem>PC</sourcesystem><sourcerecordid>2102_00700</sourcerecordid><originalsourceid>FETCH-LOGICAL-a670-c107597019edc318b4f140c9e99cd589c9b4aab055fb7a94f695c6cee428107b3</originalsourceid><addsrcrecordid>eNotj71OwzAYRb0woMIDMGF1T7ETO4nHKJSCVMFA98h2PicW-ZPj0ObtMYXpLOde6SD0QMmO5ZyTJ-ku9nsXUxLvCMkIuUVLgR1MbqwXbZXtrF_x7Jd6xaPB22Jpehi8HRp8gAG81bjomtFZ3_YzPgfgZ4AJv8PiZBfgz6P7mrEZHd5fpi6YYepbwGULvdXB-Zykhu0dujGym-H-nxt0etmfytfo-HF4K4tjJNOMRJqSjIuMUAG1TmiumKGMaAFC6JrnQgvFpFSEc6MyKZhJBdepBmBxHqYq2aDHv9trdzU520u3Vr_91bU_-QGqIFbb</addsrcrecordid><sourcetype>Open Access Repository</sourcetype><iscdi>true</iscdi><recordtype>article</recordtype></control><display><type>article</type><title>A reproducibility study of "Augmenting Genetic Algorithms with Deep Neural Networks for Exploring the Chemical Space"</title><source>arXiv.org</source><creator>Jablonka, Kevin Maik ; Mcilwaine, Fergus ; Garcia, Susana ; Smit, Berend ; Yoo, Brian</creator><creatorcontrib>Jablonka, Kevin Maik ; Mcilwaine, Fergus ; Garcia, Susana ; Smit, Berend ; Yoo, Brian</creatorcontrib><description>Nigam et al. reported a genetic algorithm (GA) utilizing the SELFIES representation and also propose an adaptive, neural network-based penalty that is supposed to improve the diversity of the generated molecules. The main claims of the paper are that this GA outperforms other generative techniques (as measured by the penalized logP) and that a neural network-based adaptive penalty increases the diversity of the generated molecules. In this work, we investigated the reproducibility of their claims. Overall, we were able to reproduce comparable results using the SELFIES-based GA, but mostly by exploiting deficiencies of the (easily optimizable) fitness function (i.e., generating long, sulfur containing chains). In addition, we reproduce results showing that the discriminator can be used to bias the generation of molecules to ones that are similar to the reference set. Lastly, we attempted to quantify the evolution of the diversity, understand the influence of some hyperparameters, and propose improvements to the adaptive penalty.</description><identifier>DOI: 10.48550/arxiv.2102.00700</identifier><language>eng</language><subject>Computer Science - Learning ; Physics - Materials Science</subject><creationdate>2021-02</creationdate><rights>http://arxiv.org/licenses/nonexclusive-distrib/1.0</rights><oa>free_for_read</oa><woscitedreferencessubscribed>false</woscitedreferencessubscribed></display><links><openurl>$$Topenurl_article</openurl><openurlfulltext>$$Topenurlfull_article</openurlfulltext><thumbnail>$$Tsyndetics_thumb_exl</thumbnail><link.rule.ids>228,230,780,885</link.rule.ids><linktorsrc>$$Uhttps://arxiv.org/abs/2102.00700$$EView_record_in_Cornell_University$$FView_record_in_$$GCornell_University$$Hfree_for_read</linktorsrc><backlink>$$Uhttps://doi.org/10.48550/arXiv.2102.00700$$DView paper in arXiv$$Hfree_for_read</backlink></links><search><creatorcontrib>Jablonka, Kevin Maik</creatorcontrib><creatorcontrib>Mcilwaine, Fergus</creatorcontrib><creatorcontrib>Garcia, Susana</creatorcontrib><creatorcontrib>Smit, Berend</creatorcontrib><creatorcontrib>Yoo, Brian</creatorcontrib><title>A reproducibility study of "Augmenting Genetic Algorithms with Deep Neural Networks for Exploring the Chemical Space"</title><description>Nigam et al. reported a genetic algorithm (GA) utilizing the SELFIES representation and also propose an adaptive, neural network-based penalty that is supposed to improve the diversity of the generated molecules. The main claims of the paper are that this GA outperforms other generative techniques (as measured by the penalized logP) and that a neural network-based adaptive penalty increases the diversity of the generated molecules. In this work, we investigated the reproducibility of their claims. Overall, we were able to reproduce comparable results using the SELFIES-based GA, but mostly by exploiting deficiencies of the (easily optimizable) fitness function (i.e., generating long, sulfur containing chains). In addition, we reproduce results showing that the discriminator can be used to bias the generation of molecules to ones that are similar to the reference set. Lastly, we attempted to quantify the evolution of the diversity, understand the influence of some hyperparameters, and propose improvements to the adaptive penalty.</description><subject>Computer Science - Learning</subject><subject>Physics - Materials Science</subject><fulltext>true</fulltext><rsrctype>article</rsrctype><creationdate>2021</creationdate><recordtype>article</recordtype><sourceid>GOX</sourceid><recordid>eNotj71OwzAYRb0woMIDMGF1T7ETO4nHKJSCVMFA98h2PicW-ZPj0ObtMYXpLOde6SD0QMmO5ZyTJ-ku9nsXUxLvCMkIuUVLgR1MbqwXbZXtrF_x7Jd6xaPB22Jpehi8HRp8gAG81bjomtFZ3_YzPgfgZ4AJv8PiZBfgz6P7mrEZHd5fpi6YYepbwGULvdXB-Zykhu0dujGym-H-nxt0etmfytfo-HF4K4tjJNOMRJqSjIuMUAG1TmiumKGMaAFC6JrnQgvFpFSEc6MyKZhJBdepBmBxHqYq2aDHv9trdzU520u3Vr_91bU_-QGqIFbb</recordid><startdate>20210201</startdate><enddate>20210201</enddate><creator>Jablonka, Kevin Maik</creator><creator>Mcilwaine, Fergus</creator><creator>Garcia, Susana</creator><creator>Smit, Berend</creator><creator>Yoo, Brian</creator><scope>AKY</scope><scope>GOX</scope></search><sort><creationdate>20210201</creationdate><title>A reproducibility study of "Augmenting Genetic Algorithms with Deep Neural Networks for Exploring the Chemical Space"</title><author>Jablonka, Kevin Maik ; Mcilwaine, Fergus ; Garcia, Susana ; Smit, Berend ; Yoo, Brian</author></sort><facets><frbrtype>5</frbrtype><frbrgroupid>cdi_FETCH-LOGICAL-a670-c107597019edc318b4f140c9e99cd589c9b4aab055fb7a94f695c6cee428107b3</frbrgroupid><rsrctype>articles</rsrctype><prefilter>articles</prefilter><language>eng</language><creationdate>2021</creationdate><topic>Computer Science - Learning</topic><topic>Physics - Materials Science</topic><toplevel>online_resources</toplevel><creatorcontrib>Jablonka, Kevin Maik</creatorcontrib><creatorcontrib>Mcilwaine, Fergus</creatorcontrib><creatorcontrib>Garcia, Susana</creatorcontrib><creatorcontrib>Smit, Berend</creatorcontrib><creatorcontrib>Yoo, Brian</creatorcontrib><collection>arXiv Computer Science</collection><collection>arXiv.org</collection></facets><delivery><delcategory>Remote Search Resource</delcategory><fulltext>fulltext_linktorsrc</fulltext></delivery><addata><au>Jablonka, Kevin Maik</au><au>Mcilwaine, Fergus</au><au>Garcia, Susana</au><au>Smit, Berend</au><au>Yoo, Brian</au><format>journal</format><genre>article</genre><ristype>JOUR</ristype><atitle>A reproducibility study of "Augmenting Genetic Algorithms with Deep Neural Networks for Exploring the Chemical Space"</atitle><date>2021-02-01</date><risdate>2021</risdate><abstract>Nigam et al. reported a genetic algorithm (GA) utilizing the SELFIES representation and also propose an adaptive, neural network-based penalty that is supposed to improve the diversity of the generated molecules. The main claims of the paper are that this GA outperforms other generative techniques (as measured by the penalized logP) and that a neural network-based adaptive penalty increases the diversity of the generated molecules. In this work, we investigated the reproducibility of their claims. Overall, we were able to reproduce comparable results using the SELFIES-based GA, but mostly by exploiting deficiencies of the (easily optimizable) fitness function (i.e., generating long, sulfur containing chains). In addition, we reproduce results showing that the discriminator can be used to bias the generation of molecules to ones that are similar to the reference set. Lastly, we attempted to quantify the evolution of the diversity, understand the influence of some hyperparameters, and propose improvements to the adaptive penalty.</abstract><doi>10.48550/arxiv.2102.00700</doi><oa>free_for_read</oa></addata></record>
fulltext fulltext_linktorsrc
identifier DOI: 10.48550/arxiv.2102.00700
ispartof
issn
language eng
recordid cdi_arxiv_primary_2102_00700
source arXiv.org
subjects Computer Science - Learning
Physics - Materials Science
title A reproducibility study of "Augmenting Genetic Algorithms with Deep Neural Networks for Exploring the Chemical Space"
url https://sfx.bib-bvb.de/sfx_tum?ctx_ver=Z39.88-2004&ctx_enc=info:ofi/enc:UTF-8&ctx_tim=2025-01-13T20%3A09%3A35IST&url_ver=Z39.88-2004&url_ctx_fmt=infofi/fmt:kev:mtx:ctx&rfr_id=info:sid/primo.exlibrisgroup.com:primo3-Article-arxiv_GOX&rft_val_fmt=info:ofi/fmt:kev:mtx:journal&rft.genre=article&rft.atitle=A%20reproducibility%20study%20of%20%22Augmenting%20Genetic%20Algorithms%20with%20Deep%20Neural%20Networks%20for%20Exploring%20the%20Chemical%20Space%22&rft.au=Jablonka,%20Kevin%20Maik&rft.date=2021-02-01&rft_id=info:doi/10.48550/arxiv.2102.00700&rft_dat=%3Carxiv_GOX%3E2102_00700%3C/arxiv_GOX%3E%3Curl%3E%3C/url%3E&disable_directlink=true&sfx.directlink=off&sfx.report_link=0&rft_id=info:oai/&rft_id=info:pmid/&rfr_iscdi=true