Matej Balog

I'm working as a Research Scientist on the Science team at DeepMind in London. Previously I was a PhD student on the Cambridge-Tübingen Machine Learning PhD Fellowship, jointly supervised by Prof Zoubin Ghahramani at the Department of Engineering, University of Cambridge and by Prof Bernhard Schölkopf at the Max Planck Institute for Intelligent Systems in Tübingen. I had graduated from Merton College, University of Oxford with the degree Master of Mathematics and Computer Science.

Converting to Optimization in Machine Learning: Perturb-and-MAP, Differential Privacy, and Program Synthesis

Matej Balog
PhD thesis

PDF

Neural Program Synthesis with a Differentiable Fixer

Matej Balog, Rishabh Singh, Petros Maniatis, Charles Sutton

Paper (PDF)

arXiv (link)

Fast Training of Sparse Graph Neural Networks on Dense Hardware

Matej Balog, Bart van Merriënboer, Subhodeep Moitra, Yujia Li, Daniel Tarlow

Paper (PDF)

arXiv (link)

Differentially Private Database Release via Kernel Mean Embeddings

Matej Balog, Ilya Tolstikhin, Bernhard Schölkopf
35th International Conference on Machine Learning (ICML 2018)

Paper (PDF)

arXiv (link)

Code on GitHub (link)

BibTeX

Lost Relatives of the Gumbel Trick

Matej Balog, Nilesh Tripuraneni, Zoubin Ghahramani, Adrian Weller
34th International Conference on Machine Learning (ICML 2017)
(Best Paper Honourable Mention)

Paper (PDF)

arXiv (link)

Poster (PDF)

Slides (PDF)

Code on GitHub (link)

BibTeX

DeepCoder: Learning to Write Programs

Matej Balog, Alexander L. Gaunt, Marc Brockschmidt, Sebastian Nowozin, Daniel Tarlow
5th International Conference on Learning Representations (ICLR 2017)
(Media coverage: smerity.com; New Scientist; WIRED)

Paper (PDF)

OpenReview (link)

arXiv (link)

BibTeX

The Mondrian Kernel

Matej Balog, Balaji Lakshminarayanan, Zoubin Ghahramani, Daniel M. Roy, Yee Whye Teh
32nd Conference on Uncertainty in Artificial Intelligence (UAI 2016)
(Oral presentation)

Paper (PDF)

Supplement (PDF)

arXiv (link)

Poster (PDF)

Slides (PDF)

Code on GitHub (link)

BibTeX

The Mondrian Process in Machine Learning (Part C project)

Matej Balog, Yee Whye Teh
Part C project (4-th year project), University of Oxford, 2015

My 4-th year project focused on applications of the Mondrian process in machine learning. In particular, efficient random feature approximations of the Laplace kernel were investigated.

Thesis (PDF)

Poster (PDF)

arXiv (link)