About me #

I’m a postdoc at Vector Institute researching generative modeling, AI applications in natural sciences, MCMC with Alireza Makhzani. Previously I was a postdoc at the University of Amsterdam with Max Welling.

I was born in Sevastopol (Ukraine) and at the age of 13 I started competing in Ukrainian olympiads on math and physics. Since then I’m interested in studying and understanding nature in all its forms from abstract fields (math, physics, computer science, which I have been studying my whole life) to the practical fields (history, biology, humanities, which I study from pop-science sources in my free time).

Selected Papers #

• Action Matching
  Kirill Neklyudov, Rob Brekelmans, Daniel Severo, Alireza Makhzani
  [arXiv] [github] [talk]

MCMC #

• Orbital MCMC (AISTATS 2022, oral)
  Kirill Neklyudov, Max Welling
  [arXiv] [github]
• Involutive MCMC: a Unifying Framework (ICML 2020)
  Kirill Neklyudov, Max Welling, Evgenii Egorov, Dmitry Vetrov
  [arXiv] [github]
• Deterministic Gibbs Sampling via ODEs
  Kirill Neklyudov, Roberto Bondesan, Max Welling
  [arXiv] [github]
• The Implicit Metropolis-Hastings Algorithm (NeurIPS 2019)
  Kirill Neklyudov, Evgenii Egorov, Dmitry Vetrov
  [arXiv]

Bayesian Deep Learning #

• Variance Networks (ICLR 2019)
  {Kirill Neklyudov, Dmitry Molchanov, Arsenii Ashukha}^*, Dmitry Vetrov
  [arXiv] [github]
• Structured Bayesian Pruning (NeurIPS 2017)
  Kirill Neklyudov, Dmitry Molchanov, Arsenii Ashukha, Dmitry Vetrov
  [arXiv] [github]

* (joint main-authorship)

Talks #

• Action Matching talk at Shannon’s Bandwagon, 2023
• Langevin dynamics for sampling and global optimization, 2019
• Sparse variational dropout and variance networks, 2018

Email #

You can find my email in my papers.