Elizaveta Semenova

Elizaveta Semenova

Lecturer in Biostatistics, Computational Epidemiology and Machine Learning

Imperial College London

Biography

I am a lecturer in Biostatistics, Computational Epidemiology and Machine Learning at Imperial College London, Department of Epidemiology and Biostatistics. And I also hold Schmidt Sciences AI2050 Early Career Fellowship.

Research

My work is centered around scalable and flexible methods for spatiotemporal statistics and Bayesian machine learning with applications in epidemiology. Most recently, my focus has been on using deep generative modelling to power MCMC inference in classical spatial statistics (the same methods work disease transmission modelling too!). Also of interest is adaptive survey design.

Main outputs of the DGMs for MCMC theme (please cite PriorCVAE if you are using the method):

  • “PriorVAE: encoding spatial priors with variational autoencoders for small-area estimation”, Semenova, Xu, Howes, Rashid, Bhatt, Mishra, Flaxman

  • “Deep learning and MCMC with aggVAE for shifting administrative boundaries: mapping malaria prevalence in Kenya”, Semenova, Mishra, Bhatt, Flaxman, Unwin

  • “PriorCVAE: scalable MCMC parameter inference with Bayesian deep generative modelling”, Semenova, Verma, Cairney-Leeming, Solin, Bhatt, Flaxman

Previously I worked at the University of Oxford, Computer Science (2022-2024) and Imperial College London, Department of Mathematics, Statistics section (2021-2022) with Seth Flaxman and MLGH network. Before that I did a postdoc in Bayesian Machine Learning at AstraZeneca R&D (2019-2021) where I also collaborated with Prioris.ai.

In 2019 I completed a PhD in Epidemiology at the Swiss TPH, where I worked on modelling of point pattern data using Log-Gaussian Cox Process and detection of hotspots on gridded surfaces.

Service

My most recent organisational activities include (1). StanCon 2024; (2). ICLR'23 “First workshop on Machine Learning & Global Health”; (3). NeurIPS'22 “Workshop on Gaussian Processes, Spatiotemporal Modeling, and Decision-making Systems”; (4). Gaussian Processes seminar series; (5). Data Science Theme Ambassador at Imperial College London.

Teaching

In spring 2024 I had an opportunity to teach a course “Bayesian Modelling and Probabilistic Programming with Numpyro” for the “AI for Science” MSc at the African Institute for Mathematical Sciences, South Africa. Lecture notes (some to be finalised) are available online.

Interests
  • Spatiotemporal statistics
  • Gaussian processes
  • Deep generative models
  • Bayesian survey design
  • Epidemiological applications
Education
  • PhD (summa cum laude) in Epidemiology, 2019

    Swiss Tropical and Public Health Institute (TPH), University of Basel, Switzerland

  • Diploma (first class honours) in Mathematics, 2008

    Moscow State University, Russia