[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-detail-119501-en":3,"doc-seo-119501-105":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},119501,962075114101,"Seraphina","https://ap-avatar.wpscdn.com/avatar/e000253a75eb197efd?x-image-process=image/resize,m_fixed,w_180,h_180&k=1780044092746381165",8,"Research & Report","Generative Models for Generic Data - Thesis","Generative models parameterized by large neural networks have become widespread, enabling realistic generation across domains such as text and chemical structures. As applications expand, models must be adapted to the heterogeneous and structured data encountered in real-world settings. This thesis develops generative frameworks that operate on generic data and accommodate diverse structures and regularities. The work studies iterative generative processes, then introduces CTMC-based discrete generation, dimension-adaptive diffusion, and flow-driven modeling for protein motif scaffolding and co-design.","Generative Models for Generic Data  \nAndrew Campbell St Peter’s College  \nUniversity of Oxford  \nA thesis submitted for the degree of Doctor of Philosophy  \nHilary 2025  \nAcknowledgements  \nFirst and foremost I would like to thank my advisors Arnaud and Tom. I greatly appreciate their belief in me and unwavering support. There were countless times during my PhD where I may have been at a loss as to what to do but our meetings always gave me a new confidence and excitement for research.  \nI would like to thank all the friends made over these past 4 years for making this such a great experience, among them are Adam F, Adam G, Amitis, Angus, Anna, Bobby, Carlo, Chris, Dan, Desi, Emi, Emile, Faaiz, Fabian, Freddie, Guneet, Gonzalo, Jakiw, James, Jannik, Jef, Jin, Justin, Kamélia, Leo, Marcus, Martin, Max, Michael, Rob, Robert, Sahra, Saif, Sam, Shahine, Thu, Tim, Tyler, Valentin, Vik and Yuyang. I’d also like to thank the department staff for all their help, especially Frédérique, Joanna, Jonathan, Mark and Stuart for always helping me out with any question I had.  \nI owe a huge debt of gratitude to the wonderful collaborators I have had the opportunity to work including Vincent, Wenlong, Yuyang, Joe, Valentin, Will, Christian, Jason, Emile, Andrew, Chris, Saif and Sulin. I have learnt so much from working together and this thesis would not have been possible without them.  \nI am also truly indebted to my internship mentors Adi at Tiktok, José at MSR and Arnaud at GDM for giving me the opportunity to work on exciting problems and welcoming me to the team. I have fond memories of the lunches and table tennis with the AI4Science team in Cambridge and the regular breakfastsand lunches with friends at GDM.  \nI owe a big thank you to Jason and Ignacio for bringing about my research visit and Tommi and Regina for hosting me. I learned a great deal from my time in Cambridge and would like to thank everyone at CSAIL who made me feel welcomed including Bowen, Gabri, Hannes, Itamar, Jakub, Jeremy, Mingyu, Peter H, Peter M, Sean, Sulin, Tally, Timur, Xiang and Yilun.  \nFinally, and most dear to me, I must give my deepest thanks to my mum Gill, my dad Paul and my sister Charlotte. Without your constant love and support I would not have been able to reach the end of this journey. Thank you for always being there for me.  \nAbstract  \nGenerative models parameterized through large neural networks are now ubiquitous across machine learning. These models are able to generate highly realistic samples, ranging from human-like text to novel chemical compounds. With the ever growing list of potential application areas comes the need to continually adapt generative models to the complex data types that appear in the real world. This thesis aims to endow generative models with the ability to operate over generic data, requiring our generative frameworks to be general enough to handle any structures and regularities that may be present in the dataset. We focus on the subfamily of generative models that create data by simulating an iterative generative process which have become the best performing model class across a range of tasks due to their simple training procedure and high quality samples. This brings new challenges for generic data modelling because appropriate generative processes must be designed to operate on complex data spaces.  \nWe begin by developing a method for generating discrete data through simulating continuous time Markov chains (CTMCs) . Our approach translates techniques from continuous space diffusion models parameterized by stochastic differential equations into discrete space. We then turn our attention to data that can vary in dimensionality by giving continuous space diffusion models the ability to add dimensions as needed during generation. In the second half of the thesis, we switch focus to flow-based generative models which offer a simple alternative approach to creating generative processes. We first apply flows operating ","cbCairyz1DWJUVzX","https://ap.wps.com/l/cbCairyz1DWJUVzX","pdf",23470192,1,245,"English","en",105,"# Introduction\n## Why is Generative Modelling Difficult?\n## Variational Autoencoders\n## Generative Adversarial Networks\n## Multi-Stage Generative Models\n## From Generative Models to Generative Processes\n## Stochastic Processes\n## Diffusion and Flow-Based Models\n## Thesis Outline\n# Literature Review\n## Continuous Space Diffusion Models\n## Discrete Space Diffusion Models\n## Flow-Based Models","[{\"question\":\"What does the thesis aim to achieve in generative modeling?\",\"answer\":\"It aims to equip generative models to operate over generic data, with frameworks general enough to handle varying structures and regularities present in datasets.\"},{\"question\":\"How does the thesis generate discrete data?\",\"answer\":\"It develops a method that simulates continuous time Markov chains (CTMCs) and translates ideas from continuous-space diffusion models parameterized by stochastic differential equations into discrete space.\"},{\"question\":\"What flow-based tasks are explored in the thesis?\",\"answer\":\"It applies flows on the space of rigid body motions to protein motif-scaffolding, then extends flow-based modeling to discrete and multi-modal data using CTMCs combined with continuous-space flows for protein co-design.\"}]","Generative Models for Generic Data - Thesis | PDF",1785724677,617,{"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},"generative-models-for-generic-data-thesis","",{"@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/generative-models-for-generic-data-thesis/119501/",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","2026-08-03",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 does the thesis aim to achieve in generative modeling?","Question",{"text":75,"@type":76},"It aims to equip generative models to operate over generic data, with frameworks general enough to handle varying structures and regularities present in datasets.","Answer",{"name":78,"@type":73,"acceptedAnswer":79},"How does the thesis generate discrete data?",{"text":80,"@type":76},"It develops a method that simulates continuous time Markov chains (CTMCs) and translates ideas from continuous-space diffusion models parameterized by stochastic differential equations into discrete space.",{"name":82,"@type":73,"acceptedAnswer":83},"What flow-based tasks are explored in the thesis?",{"text":84,"@type":76},"It applies flows on the space of rigid body motions to protein motif-scaffolding, then extends flow-based modeling to discrete and multi-modal data using CTMCs combined with continuous-space flows for protein co-design.","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"]