Dr. Stefan Radev

Dr. Stefan Radev

Dr. Stefan Radev

STRUCTURES Cluster of Excellence
Berliner Str. 47
69120 Heidelberg

SMiP Advisors:  Prof. Dr. Andreas Voss, Prof. Dr. Thorsten Meiser, Prof. Jeffrey Rouder, Ph.D.

Thesis:  Deep learning architectures for amortized Bayesian inference in cognitive modeling

Degree: Dr. phil. (summa cum laude), received from the University of Heidelberg in May 2021

First placement: Start-up funding from RTG SMiP, researcher at the University of Heidelberg

Current position: Postdoctoral researcher at STRUCTURES Cluster of Excellence, Heidelberg University

  • Research Areas

    • Machine learning & deep learning
    • Bayesian analysis
    • Simulation science
    • Cognitive modeling
  • Teaching (SMIP)

    • Workshop „Python Basics“ (together with Ulf Mertens)
  • Publications

    von Krause*, M., Radev*, S. T., & Voss, A. (2022). Mental speed is high until age 60 as revealed by analysis of over a million participants. Nature human behaviour6(5), 700–708. https://doi.org/10.1038/s41562–021-01282–7 [*shared first authorship]

    Radev, S. T., Graw, F., Chen, S., Mutters, N. T., Eichel, V. M., Bärnighausen, T., & Köthe, U. (2021). OutbreakFlow: Model-based Bayesian inference of disease outbreak dynamics with invertible neural networks and its application to the COVID-19 pandemics in Germany. PLoS computational biology17(10), e1009472. https://doi.org/10.1371/journal.pcbi.1009472

    Bieringer, S., Butter, A., Heimel, T., Höche, S., Köthe, U., Plehn, T., & Radev, S. T. (2021). Measuring QCD splittings with invertible networks. SciPost Physics10(6), 126. https://doi.org/10.21468/SciPostPhys.10.6.126 

    Radev, S. T., D'Alessandro, M., Mertens, U. K., Voss, A., Köthe, U., & Bürkner, P. C. (2021). Amortized Bayesian model comparison with evidential deep learning. IEEE Transactions on Neural Networks and Learning Systems. https://doi.org/10.1109/TNNLS.2021.3124052 

    Radev, S. T., Mertens, U. K., Voss, A., Ardizzone, L., & Köthe, U. (2020). BayesFlow: Learning complex stochastic models with invertible neural networks. IEEE Transactions on Neural Networks and Learning systems33(4), 1452–1466. https://doi.org/10.1109/TNNLS.2020.3042395 

    all publications

  • Talks

    Radev, S. (2019, September). Pushing the boundaries: Tackling intractable response-time models with deep learning methods [Talk]. 14th Conference of the Section Methods and Evaluation of the German Psychological Society (DGPs). Kiel, Germany.

  • Posters

    Radev, S.T. (2019, June). Taming the Intractable: Deep Learning for Universal Parameter Estimation. Poster at the 34th IOPS/SMiP Summer Conference, 13–14 June 2019, Utrecht, Netherlands.

    Radev, S.T. (2019). BayesFlow: Learning complex stochastic models with invertible neural networks. Poster presented at the 50th Meeting of the European Mathematical Psychology Group (EMPG), Heidelberg, Germany.

    Radev, S.T., Mertens, U., Voss, A. (2018). Abrox – a graphical user interface for approximate Bayesian computation. In: 60. Tagung experimentell arbeitender Psychologen. Marburg, Germany.

    Radev, S. T., Lerche, V., Mertens, U., & Voss, A. (July 2017). Diffusion model analysis: a graphical user interface with fast-dm. Poster presented at the „50th Annual Meeting of the Society for Mathematical Psychology (MathPsych)“ 2017 in Warwick.

    Radev, S. T., Lerche, V., Mertens, U., & Voss, A. (September 2017). Diffusion model analysis: a graphical user interface to fast-dm. Poster presented at the „13. Tagung der Fach­gruppe Methoden & Evaluation der Deutschen Gesellschaft für Psychologie (FGME)“ 2017 in Tübingen.