[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-detail-118570-en":3,"doc-seo-118570-105":30,"detail-sidebar-cat-0-en-105":91},{"code":4,"msg":5,"data":6},0,"success",{"doc_id":7,"user_id":8,"nickname":9,"user_avatar":10,"doc_module":4,"category_id":11,"category_name":12,"doc_title":13,"doc_description":14,"doc_content":15,"file_id":16,"file_url":17,"file_type":18,"file_size":19,"view_count":4,"is_deleted":4,"is_public":20,"is_downloadable":20,"audit_status":20,"page_count":21,"language":22,"language_code":23,"site_id":24,"html_lang":23,"table_of_contents":25,"faqs":26,"seo_title":27,"seo_description":14,"update_tm":28,"read_time":29},118570,1374391974564,"Clementine","https://ap-avatar.wpscdn.com/avatar/14000253aa45c000a9e?x-image-process=image/resize,m_fixed,w_180,h_180&k=1779874745381141002",8,"Research & Report","Kinetic predictions for E2 and SN2 reactions using the BERT architecture - Comparison and interpretation","Accurate prediction of reaction rates supports reaction mechanism elucidation and the design of synthetic pathways. This thesis trains a BERT-based transformer model to predict experimental log values for E2 and SN2 reactions from SMILES representations, then interprets its outputs using established reactivity rules. Performance is first benchmarked against a Random Forest baseline for SN2 kinetics. The model is subsequently fine-tuned for joint E2/SN2 prediction, achieving near-experimental accuracy and highlighting key structural and physical effects, enabling interpretable guidance for synthesis.","Kinetic predictions for E2 and SN 2 reactions using the BERT architecture: Comparison  \nand interpretation  \nChloe A. Wilson Lady Margaret Hall University of Oxford  \nA thesis submitted for the degree of Doctor of Philosophy  \nMichaelmas 2024  \nAbstract  \nThe accurate prediction of reaction rates is an integral step in reaction mechanism elucidation and design of synthetic pathways. Traditionally, kinetic parameters have been derived from activation energies obtained from Quantum Mechanics (QM) methods, and more recently, machine learning (ML) . Transformer-based models have emerged as a popular ML method for chemical prediction, allowing reactions to be represented as SMILES strings. Despite being state-of-theart in both reaction classification and yield prediction, Bidirectional Encoder Representations from Transformers (BERT), a type of transformer-based model, have not yet been applied to kinetic prediction. In this Thesis, a BERT model is trained to predict experimental log 􀀺 values of E2 and SN2 reactions, and its predictions are interpreted in the context of known reactivity rules. Initially, BERT is trained to predict log 􀀺 of SN2 reactions and its performance is compared to the state-of-the-art Random Forest (RF) model in terms of accuracy, training time, and recognition of key structural and physical effects (Chapter 3) . This model is subsequently fine-tuned to predict log 􀀺 of both E2 and SN2 reactions, and its predictions are interpreted regarding known drivers of mechanistic competition (Chapter 4) . The final model exhibits strong performance, achieving nearexperimental accuracy and identifying key structural and physical effects. Overall, this Thesis presents a framework for the accurate and interpretable prediction of reaction kinetics, with potential applications in guiding synthetic design.  \nDeclaration  \nI, Chloe Wilson, declare that the Thesis I am submitting is entirely my own work except where indicated in the text, caption, footnote, or bibliography. This Thesis has not been submitted in whole or in part for any other academic degree or professional qualification.  \n27th November 2024  \nSignature Date  \nAcknowledgements  \nI would like to thank my supervisors, Profs. Fernanda Duarte and Jason Crain, for their guidance and support throughout my DPhil, and for giving me the opportunity to work on this project. The skills I have developed under Fernanda and Jason’s mentorship are invaluable, and I will be forever grateful to them for their belief in me. To Fernanda, thank you for welcoming me into your research group and for helping me to grow as a scientist. Fernanda taught me how to shape my ideas and findings into a cohesive project, and consistently encouraged me to refine my work. To Jason, thank you for taking the time to provide valuable technical insights. Jason’s role in introducing me to the BERT architecture and IBM Cloud platform have been instrumental in the progression of this project.  \nI cannot thank my supervisors without also acknowledging my unofficial supervisors: Matina, Ewa, Tom W, and James -past and present members of the Duarte group - without whom, none of this work would have been possible. Matina, for always being just a video call away whenever I needed her. Ewa and Tom, for staying behind after my presentations to offer incredibly detailed (and useful!) advice. And James, for his abundance of scientific knowledge, which he has always kindly shared.  \nAdditional thanks toMatina and Tom W, as well as previous Duarte group members Ali and Tom Y, for their incredible help during my first year of DPhil. They were patient with me while I got to grips with the technical aspects of the project. I am especially grateful for their support during the pandemic, as many of them had not yet met me in person, but still took the time to advise me over video call.  \nI would also like to thank María Calvo, whom I supervised during her summer project in the Duarte group. The results from the QM ","cbCaitJmnzU5BJGh","https://ap.wps.com/l/cbCaitJmnzU5BJGh","pdf",11671236,1,252,"English","en",105,"# Abstract\n# Declaration\n# Acknowledgements\n# Data availability","[{\"question\":\"What problem does the thesis address?\",\"answer\":\"It addresses accurate prediction of reaction rates to support reaction mechanism elucidation and synthetic pathway design.\"},{\"question\":\"How is the BERT model used in this work?\",\"answer\":\"A BERT model is trained to predict experimental log values for E2 and SN2 reactions from reaction representations, then its predictions are interpreted using known reactivity rules.\"},{\"question\":\"What comparisons and evaluations are performed?\",\"answer\":\"For SN2, the BERT model’s performance is compared with a Random Forest model using measures such as accuracy, training time, and recognition of key structural and physical effects.\"}]","Kinetic predictions for E2 and SN2 reactions using the BERT architecture - Comparison and interpretation | PDF",1785684294,635,{"code":4,"msg":31,"data":32},"ok",{"site_id":24,"language":23,"slug":33,"title":13,"keywords":34,"description":14,"schema_data":35,"social_meta":86,"head_meta":88,"extra_data":90,"updated_unix":28},"kinetic-predictions-for-e2-and-sn2-reactions-using-the-bert-architecture-comparison-and-interpretation","",{"@graph":36,"@context":85},[37,54,68],{"@type":38,"itemListElement":39},"BreadcrumbList",[40,44,48,51],{"item":41,"name":42,"@type":43,"position":20},"https://docshare.wps.com","Home","ListItem",{"item":45,"name":46,"@type":43,"position":47},"https://docshare.wps.com/document/","Document",2,{"item":49,"name":12,"@type":43,"position":50},"https://docshare.wps.com/document/research-report/",3,{"item":52,"name":13,"@type":43,"position":53},"https://docshare.wps.com/document/kinetic-predictions-for-e2-and-sn2-reactions-using-the-bert-architecture-comparison-and-interpretation/118570/",4,{"url":52,"name":13,"@type":55,"author":56,"headline":13,"publisher":58,"fileFormat":61,"inLanguage":23,"description":14,"dateModified":62,"datePublished":62,"encodingFormat":61,"isAccessibleForFree":63,"interactionStatistic":64},"DigitalDocument",{"name":9,"@type":57},"Person",{"url":41,"name":59,"@type":60},"DocShare","Organization","application/pdf","2026-08-02",true,{"@type":65,"interactionType":66,"userInteractionCount":4},"InteractionCounter",{"@type":67},"ViewAction",{"@type":69,"mainEntity":70},"FAQPage",[71,77,81],{"name":72,"@type":73,"acceptedAnswer":74},"What problem does the thesis address?","Question",{"text":75,"@type":76},"It addresses accurate prediction of reaction rates to support reaction mechanism elucidation and synthetic pathway design.","Answer",{"name":78,"@type":73,"acceptedAnswer":79},"How is the BERT model used in this work?",{"text":80,"@type":76},"A BERT model is trained to predict experimental log values for E2 and SN2 reactions from reaction representations, then its predictions are interpreted using known reactivity rules.",{"name":82,"@type":73,"acceptedAnswer":83},"What comparisons and evaluations are performed?",{"text":84,"@type":76},"For SN2, the BERT model’s performance is compared with a Random Forest model using measures such as accuracy, training time, and recognition of key structural and physical effects.","https://schema.org",{"og:url":52,"og:type":87,"og:title":13,"og:site_name":59,"og:description":14},"article",{"robots":89,"canonical":52},"index,follow",{"doc_id":7,"site_id":24},{"code":4,"msg":5,"data":92},[93,97,101,105,110,115,120,123,128,131,135],{"id":20,"doc_module":4,"doc_module_name":46,"category_name":94,"show_sort_weight":95,"slug":96},"Story & Novel",90,"story-novel",{"id":47,"doc_module":4,"doc_module_name":46,"category_name":98,"show_sort_weight":99,"slug":100},"Literature",80,"literature",{"id":53,"doc_module":4,"doc_module_name":46,"category_name":102,"show_sort_weight":103,"slug":104},"Exam",70,"exam",{"id":106,"doc_module":4,"doc_module_name":46,"category_name":107,"show_sort_weight":108,"slug":109},5,"Comic",60,"comic",{"id":111,"doc_module":4,"doc_module_name":46,"category_name":112,"show_sort_weight":113,"slug":114},6,"Technology",50,"technology",{"id":116,"doc_module":4,"doc_module_name":46,"category_name":117,"show_sort_weight":118,"slug":119},7,"Healthcare",40,"healthcare",{"id":11,"doc_module":4,"doc_module_name":46,"category_name":12,"show_sort_weight":121,"slug":122},30,"research-report",{"id":124,"doc_module":4,"doc_module_name":46,"category_name":125,"show_sort_weight":126,"slug":127},9,"Religion & Spirituality",20,"religion-spirituality",{"id":126,"doc_module":4,"doc_module_name":46,"category_name":129,"show_sort_weight":126,"slug":130},"World Cup","world-cup",{"id":132,"doc_module":4,"doc_module_name":46,"category_name":133,"show_sort_weight":132,"slug":134},10,"Lifestyle","lifestyle",{"id":136,"doc_module":4,"doc_module_name":46,"category_name":137,"show_sort_weight":106,"slug":138},19,"General","general"]