I am an Assistant Professor of Artificial Intelligence at the Eindhoven University of Technology. I am generally interested in the wide field of probabilistic modeling, as this is arguably the approach to reason under uncertain knowledge. In particular, I am interested in methods which aim to strike a sensible balance between expressiveness and tractability in probabilistic modeling, such as sum-product networks.


  • February, 2020
    I presented a Tutorial on Probabilistic Circuits at AAAI’20, together with Antonio Vergari, YooJung Choi, and Guy Van den Broeck.
  • January, 2020
    Our paper, Deep Structured Mixtures of Gaussian Processes, with Martin Trapp, Robert Peharz, Franz Pernkopf, and Carl Rasmussen has been accepted at AISTATS.
  • November, 2019
    I just started my new position of Assistant Professor of AI, at TU Eindhoven!
  • September, 2019
    Our paper, Bayesian Learning of Sum-Product Networks, with Martin Trapp, Robert Peharz, Franz Pernkopf, Hong Ge, and Zoubin Ghahramani has been accepted at NeurIPS (acceptance rate 21.18%).
  • June, 2019
    You can follow me now on twitter @ropeharz
  • June, 2019
    I was invited as speaker at the ICML’19 workshop on Tractable Probabilistic Models.
  • May 14, 2019
    Our paper, Random Sum-Product Networks: A Simple and Effective Approach to Probabilistic Deep Learning, with Robert Peharz, Antonio Vergari, Karl Stelzner, Alejandro Molina, Martin Trapp, Xiaoting Shao, Kristian Kersting, and Zoubin Ghahramani, has been accepted at UAI (acceptance rate 26.2%).
  • April 22, 2019
    Our paper, Hierarchical Decompositional Mixtures of Variational Autoencoders, with Ping Liang Tan and Robert Peharz, has been accepted at ICML (acceptance rate 22.6%).
  • April 22, 2019
    Our paper, Faster Attend-Infer-Repeat with Tractable Probabilistic Models, with Karl Stelzner, Robert Peharz, and Kristian Kersting, has been accepted at ICML (acceptance rate 22.6%).
  • November 1, 2018
    Our paper, Automatic Bayesian Density Analysis, with Antonio Vergari, Alejandro Molina, Robert Peharz, Zoubin Ghahramani, Kristian Kersting, and Isabel Valera has been accepted at AAAI (acceptance rate 16.2%).
  • October 29, 2018
    Our paper, Minimal Random Code Learning: Getting Bits Back from Compressed Model Parameters, with Marton Havasi, Robert Peharz, and Miguel Hernández-Lobato has been accepted at the NIPS workshop on Compact Deep Neural Networks with industrial applications (acceptance rate 44.8%).
  • June 9, 2018
    Our paper, Hybrid Generative-Discriminative Training of Gaussian Mixture Models, with Wolfgang Roth, Robert Peharz, Sebastian Tschiatschek, and Franz Pernkopf has been accepted at Pattern Recognition Letters.
  • April 3, 2018
    I started my Marie Skłodowska-Curie project Hybrid Learning Systems utilizing Sum-Product Networks (HYBSPN).
  • March 15, 2018
    I’m presenting Sum-Product Networks for Deep Probabilistic Modeling at CamAIML hosted by Microsoft Research, Cambridge, UK.
  • January 29, 2018
    My application for a Marie Skłodowska-Curie Individual Fellowship (granted by the European Commission, H2020) was successful!
  • November 8, 2017
    Our paper, Sum-Product Autoencoding: Encoding and Decoding Representations using Sum-Product Networks, with Antonio Vergari, Robert Peharz, Nicola Di Mauro, Alejandro Molina, Kristian Kersting, and Floriana Esposito, has been accepted at AAAI.
  • June 12, 2017
    Our paper, Safe Semi-Supervised Learning of Sum-Product Networks, with Martin Trapp, Tamas Madl, Robert Peharz, Franz Pernkopf, and Robert Trappl, has been accepted at UAI.