{"about":{"site":"https://codewithpapers.app","non_affiliation":"Code with Papers and Syntology are not affiliated with, endorsed by, or sponsored by Papers with Code, Meta, or the pwc-archive mirror.","licence":"CC BY-SA 4.0","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","attribution":"https://codewithpapers.app/attribution","modified":"archive material modified by Syntology; see the attribution page"},"url":"/paper/scalable-generalized-dynamic-topic-models","title":"Scalable Generalized Dynamic Topic Models","arxiv_id":"1803.07868","date":"2018-03-21","proceeding":null,"authors":["Patrick Jähnichen","Florian Wenzel","Marius Kloft","Stephan Mandt"],"abstract":"Dynamic topic models (DTMs) model the evolution of prevalent themes in\nliterature, online media, and other forms of text over time. DTMs assume that\nword co-occurrence statistics change continuously and therefore impose\ncontinuous stochastic process priors on their model parameters. These dynamical\npriors make inference much harder than in regular topic models, and also limit\nscalability. In this paper, we present several new results around DTMs. First,\nwe extend the class of tractable priors from Wiener processes to the generic\nclass of Gaussian processes (GPs). This allows us to explore topics that\ndevelop smoothly over time, that have a long-term memory or are temporally\nconcentrated (for event detection). Second, we show how to perform scalable\napproximate inference in these models based on ideas around stochastic\nvariational inference and sparse Gaussian processes. This way we can train a\nrich family of DTMs to massive data. Our experiments on several large-scale\ndatasets show that our generalized model allows us to find interesting patterns\nthat were not accessible by previous approaches.","url_abs":"http://arxiv.org/abs/1803.07868v1","url_pdf":"http://arxiv.org/pdf/1803.07868v1.pdf","source":{"archive":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","row_kind":"abstracts"},"code_links":[{"paper_slug":"scalable-generalized-dynamic-topic-models","repo_url":"https://github.com/patrickjae/GeneralizedDynamicTopicModels.jl","is_official":1,"mentioned_in_paper":0,"mentioned_in_github":0,"framework":"tf","reach":{"status":"ok","spdx":"MIT"}}],"tasks":[{"task_slug":"event-detection","task_name":"Event Detection"},{"task_slug":"gaussian-processes","task_name":"Gaussian Processes"},{"task_slug":"topic-models","task_name":"Topic Models"},{"task_slug":"variational-inference","task_name":"Variational Inference"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1803.07868","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}