CS Seminar: Olga Isupova (University of Oxford), “Dynamic topic models for video analysis and scene understanding” 2pm, Thu. 11th April

Dear all,


You are invited to the CS Department Seminar on “Dynamic topic models for video analysis and scene understanding”, by Dr Olga Isupova from the University of Oxford.


Please note that the seminar takes place on a Thursday afternoon due to the time conflict with the staff meeting on the previous day. More details can be found below and please feel free to forward this to anyone who might be interested.


Time & venue:

Thursday 11th April 14:00-15:00, 39 BB 02


Title: Dynamic topic models for video analysis and scene understanding

Speaker: Olga Isupova, University of Oxford



Topic models find latent patterns in collections of documents or images. These models are used to provide a semantic description of high-dimensional data in a compact form of topics. In many applications, particularly in video processing, consideration of dynamic dependencies in data can further improve the descriptive power of topic models. In this talk I present a series of dynamic topic models in the context of video analysis and scene understanding starting from a parametric model and diving deeper in Bayesian nonparametric models with incorporating dynamics on data. I will demonstrate that these models can extract semantical interpretable patterns in video data in a fully unsupervised way.
Olga Isupova received the Specialist (eq. to M.Sc.) degree in applied mathematics and computer science in 2012 from Lomonosov Moscow State University, Moscow, Russia, and the Ph.D. degree in 2017 from the University of Sheffield, Sheffield, U.K. She is a research assistant in machine learning at the Department of Engineering Science, University of Oxford, Oxford, U.K. Her research interests include machine learning for disaster response and environment protection, Bayesian nonparametrics, and topic modelling.