Sebastijan Dumancic

Algorithmics group. Delft University of Technology

prof_pic.jpg

Office 4.East.220

Van Mourik Broekmanweg 6

2628 XE Delft

How can we make machines learn from as little data as possible?
How can we make machines that effectively deploy their knowledge in novel situations?

I try to answer these questions by investigating how can we use programming languages to represent what artificially intelligence agents know and how they act. In pursuing this goal, I work on program synthesis, which focuses on learning programs from examples, and probabilistic programming, which focuses on representing and reasoning with probabilistic models represented as computer programs. In both fields, representing knowledge as a program is the key feature that allows us to capture knowledge of arbitrary complexity, use it flexibly, and acquire it from little experience.

I am also interested in application of program synthesis and probabilistic programming in scientific discovery (especially geophysics), transportation, robotics and planning.

Before joining TU Delft, I was an FWO-funded postdoctoral fellow in the DTAI lab at KU Leuven working with Hendrik Blockeel and Luc De Raedt, where I also obtained my PhD. I was a visiting researcher at MIT, University of Oxford, and NEC Laboratories Europe.

news

Jan 29, 2024

I’m visiting Nathanaël Fijalkow’s group at CNRS in Bordeaux

Jan 15, 2024

I’m giving a talk on the need for a neuro-symbolic languages at the Probabilistic Learning and Reasoning mini-symposium, co-located with Emile Van Krieken’s PhD defence. I am also one of Emile’s examiners.

Jan 1, 2024

Liang joins my group as a postdoc. He will be working on probabilistic programming for coupled systems.

Feb 7, 2023

Andrew, Celine and I are giving a tutorial on inductive logic programming at AAAI’23.

Jan 12, 2023

Our RAIL lab has been funded through the ROBUST programme

Dec 15, 2022

I’m hiring a postdoc to work on probabilistic programming and simulators in geophysics! Check out the details here

Nov 1, 2022

I’m visiting Josh Tenenbaum and Tom Silver at MIT.

Oct 28, 2022

Dirk van Bokkem, a Master student I co-supervised with Neil, has won the second prize 2nd prize of the Young Talent KNVI/KIVI thesis prizes for Computer Science and Information Science.

Oct 1, 2022

Tilman Hinnerichs has joined my group.

Sep 1, 2022

Issa Hanou has started a PhD with me and Mathijs de Weerdt. Issa will be working on making rail network planning more robust through insights from program synthesis and probabilistic programming.

Sep 1, 2022

A paper on automated decision-making for greenhouse climate management, with Neil and a Master student Dirk van Bokkem, is accepted at IAAI 23.

Jun 1, 2022

Our paper on provably safe machine learning, with Kshitij and Hendrik is accepted to ECML PKDD 22.

Apr 1, 2022

I’m visiting Andrew Cropper at Oxford.

Sep 9, 2021

I have joined the Algorithmics group at the Delft University of Technology as an Assistant Professor!

Sep 1, 2021

Join Behrouz, Parisa, Alex and me on our Combining Learning and Reasoning - Programming Languages, Formalisms, and Representations workshop at AAAI 22

Jan 16, 2021

I gave a talk on what abstraction can do for program induction at the Automated Verification group, Oxford University. slides

Dec 16, 2020

I gave a talk on what abstraction can do for program induction at CoCoSci lab, MIT. slides

Dec 2, 2020

Our work on knowledge refactoring for program induction is accepted to AAAI 21.

Nov 23, 2020

Our work with Teal Partners on using automated reasoning for payroll management is accepted to IAAI 21.

Sep 26, 2020

I will be giving a tutorial on neuro-symbolic AI at AAAI 21 with Luc, Giuseppe and Robin. Stay tuned!

Jun 5, 2020

A paper on tackling noise in semi-supervised clustering accepted to ECML 2020 in Ghent, Belgium.

Apr 28, 2020

Three papers accepted at IJCAI 2020 in Kyoto, Japan.

Sep 24, 2019

I’m speaking at the Decision Analytics Closing Symposium about constraint solving with Z3

Sep 12, 2019

I’m speaking at the Aritificial Intelligence Applied in Industry symposium about our work with Skyline Communications

Aug 13, 2019

I’m honoured that my PhD thesis has been awarded a honourable mention for the EurAI Distinguished Dissertation Award 2018!

Jun 21, 2019

I’m honoured to be among the Distinguished Program Committee at IJCAI 2019!

May 9, 2019

Two papers accepted at IJCAI 2019 - Learning Relational Representations with Auto-encoding Logic Programs and A Comparative Study of Distributional and Symbolic Paradigms for Relational Learning!

Nov 22, 2018

We are organising a mini-symposium on Deep learning for complex relational data on the occasion of my public PhD defence on December 11

Sep 5, 2018

The paper DeepProbLog - Neural Probabilistic Logic Programming accepted at NIPS!

Aug 10, 2018

The paper Learning Sequence Encoders for Temporal Knowledge Graph Completion accepted at EMNLP in Brussels, Belgium!

Jul 24, 2018

The paper COBRAS-TS - A new approach to Semi-Supervised Clustering of Time Series accepted at the Discovery Science 2018 to be held in Limassol, Cyprus

Jul 13, 2018

The paper COBRAS - Interactive Clustering with Pairwise Queries accepted to IDA 2018 to be held in 's-Hertogenbosch, the Netherlands

Jun 24, 2018

The paper Auto-encoding Logic programs accepted at the NAMPI workshop at IJCAI/ICML 2018 in Stockholm

Jun 15, 2018

The paper On embeddings as an alternative paradigm for relational learning accepted at the StarAI workshop at IJCAI/ICML 2018 in Stockholm

Apr 12, 2018

I’m visiting Mathias Niepert at NEC Labs Europe

Jul 31, 2017

The paper Demystifying Relational Latent Representations won the best student paper award on ILP 2017 in Orleans, France

Jun 9, 2017

The paper Demystifying Relational Latent Representations accepted at ILP 2017 in Orleans, France

Apr 23, 2017

Papers Clustering-Based Unsupervised Relational Representation Learning with an Explicit Distributed Representation and COBRA - A Fast and Simple Method for Active Clustering with Pairwise Constraints accepted at IJCAI 2017 in Melbourne, Australia

Apr 4, 2017

The paper on An expressive dissimilarity measure for relational clustering using neighbourhood trees accepted at MLJ