Thomas Kipf

Research Scientist

Google Research

I am a Research Scientist at Google Research in the Brain Team in Amsterdam. Previously, I obtained my PhD on "Deep Learning with Graph-Structured Representations" (thesis available here) at the University of Amsterdam under the supervision of Prof. Max Welling. I am broadly interested in developing and studying machine learning models that can reason about the rich structure of the physical world and its combinatorial complexity. This includes topics in graph representation learning, object-centric learning, and causal representation learning.

News

Selected Publications


Object-centric Learning with Slot Attention

Iterative, competitive attention for object discovery

F. Locatello*, D. Weissenborn, T. Unterthiner, A. Mahendran, G. Heigold, J. Uszkoreit, A. Dosovitskiy, T. Kipf*, Object-centric Learning with Slot Attention, (NeurIPS 2020), Spotlight [Link, PDF (arXiv)], *equal contribution.


Deep Learning with Graph-Structured Representations

PhD thesis (2020), University of Amsterdam

T. N. Kipf, Deep Learning with Graph-Structured Representations [Link]


Contrastive Learning of Structured World Models

Unsupervised discovery of objects, relations, and consequences of actions.

T. Kipf, E. van der Pol, M. Welling, Contrastive Learning of Structured World Models, (ICLR 2020), Oral [Link, PDF (arXiv)].


CompILE: Compositional Imitation Learning and Execution

Unsupervised, differentiable sequence segmentation for option discovery.

T. Kipf, Y. Li, H. Dai, V. Zambaldi, A. Sanchez-Gonzales, E. Grefenstette, P. Kohli, P. Battaglia, CompILE: Compositional Imitation Learning and Execution, (ICML 2019), Long Oral [Link, PDF (arXiv)].


Neural Relational Inference for Interacting Systems

Learning the latent interaction graph of a dynamical system

T. Kipf*, E. Fetaya*, K. Wang, M. Welling, R. Zemel, Neural Relational Inference for Interacting Systems, (ICML 2018) [Link, PDF (arXiv), code], *equal contribution.


MolGAN: An implicit generative model for small molecular graphs

Learning to generate molecular graphs using a combined GAN/RL-based objective

N. De Cao, T. Kipf, MolGAN: An implicit generative model for small molecular graphs, ICML Deep Generative Models Workshop (2018) [Link, PDF (arXiv), code].


Hyperspherical Variational Auto-Encoders

Learning hyperspherical latent spaces

T. R. Davidson*, L. Falorsi*, N. De Cao*, T. Kipf, J. M. Tomczak, Hyperspherical Variational Auto-Encoders, (UAI 2018), Plenary Talk [Link, PDF (arXiv), code, blog], *equal contribution.


Modeling Relational Data with Graph Convolutional Networks

Link prediction and entity classification on knowledge graphs

M. Schlichtkrull*, T. N. Kipf*, P. Bloem, R. vd Berg, I. Titov, M. Welling, Modeling Relational Data with Graph Convolutional Networks, (ESWC 2018), Best Student Research Paper [Link, PDF (arXiv), code], *equal contribution.


Semi-Supervised Classification with Graph Convolutional Networks

Neural networks for node classification on graphs

T. N. Kipf, M. Welling, Semi-Supervised Classification with Graph Convolutional Networks (ICLR 2017) [Link, PDF (arXiv), code, blog]


Variational Graph Auto-Encoders

A latent variable model for graph-structured data

T. N. Kipf, M. Welling, Variational Graph Auto-Encoders, (NeurIPS Bayesian Deep Learning Workshop 2016) [Link, PDF (arXiv), code]


For a full list, have a look at my Google Scholar page.

Curriculum vitae