Topological Analysis from Latent Semantic Analysis
Jesús Rodríguez-Viorato (CIMAT - México)
04-Feb-2022, 16:00-17:00 (2 years ago)
Abstract: Latent Semantic Analysis is one of the most widely used and accepted techniques in natural language processing. A better understanding of the topology of Latent Spaces could lead to better applications. We applied different topological techniques such as Ballmapper and persistent homology to the Latent Semantic representation of hundreds of thousands of abstracts and titles from the ArXiv database. We will present a comprehensible synthesis of our computations, comparing results between different time frames and ArXiv categories.
geometric topology
Audience: researchers in the topic
Series comments: Web-seminar series on Applications of Geometry and Topology
Organizers: | Alicia Dickenstein, José-Carlos Gómez-Larrañaga, Kathryn Hess, Neza Mramor-Kosta, Renzo Ricca*, De Witt L. Sumners |
*contact for this listing |
Export talk to