Pascal Welke

I am interested in learning on and with graphs. In particular, expressive graph representations, as well as efficient similarity based learning on graphs and graph kernel design. I wrote my PhD thesis on 'Probabilistic Frequent Subtree Mining'.

I also teach several courses that are offered by our group in the Bachelors program and Masters program in Computer Science and I supervise BA and MA theses.

My Erdős number is at most 3 (via Torsten Suel and Endre Szemerédi).

  1. Contact
  2. News
  3. Current Preprints
  4. Publications
  5. Books
  6. Non-Archival Venues
  7. Lecture Notes and Coding Nuggets
  8. Community Activities

If you like pdfs (or hate trees made out of wood), you can download my publication list in pdf format.

Contact

I have accounts on ResearchGate and LinkedIn. My publications are indexed by dblp and google scholar. Some of the code that I write is on github and I may shamelessly advertise my work and activities on twitter.

Office:
Machine Learning Research Unit
2nd Floor, Erzherzog-Johann-Platz 1 (FB02), 1040 Vienna, Austria
Email:
You can send mail to firstname.lastname@tuwien.ac.at

News

  1. Our paper Maximally Expressive GNNs for Outerplanar Graphs will soon be published in the journal Transactions on Machine Learning Research.
  2. I'll be giving a keynote talk on Expressive Graph Embeddings via Homomorphism Counting at the LoG Meetup Paris.
  3. Logical Distillation of Graph Neural Networks got an Honorable Mention Award as the best paper in the Reasoning, Learning, and Decision Making Track at KR'2024.
  4. Splitting Stump Forests received the Best Student Paper Award at the Discovery Science conference. A big shout out to Fouad Alkhoury, the first author!
  5. Our paper Weisfeiler and Leman Go Loopy: A New Hierarchy for Graph Representational Learning was accepted as an oral presentation at NeurIPS'2024. I'll bring my cactus socks to Vancouver!
  6. Is Expressivity Essential for the Predictive Performance of Graph Neural Networks? We answer No! at the Workshop on Scientific Methods for Understanding Deep Learning SciForDL@NeurIPS.
  7. Are your random forests too large, as well? Try Splitting Stump Forests to compress them without losing accuracy! Let's meet at DS'2024 in Pisa, Italy.
  8. Logical classifiers on graphs that look very similar to MPNNs? Yes! Check out our paper that got recently accepted to KR'2024 in Hanoi, Vietnam!
  9. We have organized the Graph ML social at ICML'2024 in beautiful Vienna. It was a great success with panel discussions, food, drinks, and over 200 participants.
  10. A while back, I gave an interview for "Die Presse" which was used in parts in a nice introductory article to AI.
  11. We will organize MLG@ECMLPKDD 2024, the 22th Workshop on Mining and Learning with Graphs. Join us for interesting discussions about graphs in Vilnius, Lithuania!
  12. I visited the LOG meetup in Munich. We presented our two accepted short papers Maximally Expressive GNNs for Outerplanar Graphs and Extending Graph Neural Networks with Global Features. The poster for the outerplanar GNN paper almost got the best poster award, being on second place.
  13. I am a top reviewer at NeurIPS. Awesomely, this comes with a free registration. Thank you NeurIPS!
  14. Happy that we can present two papers at the GLFrontiers@NeurIPS workshop: Graph Pooling Provably Improves Expressivity and Maximally Expressive GNNs for Outerplanar Graphs. See you in New Orleans, USA!
  15. I joined the ELLIS society as a member.
  16. We won the best poster award at G-Research's ICML poster party in London where my coauthor Max Thiessen presented Expectation-Complete Graph Representations with Homomorphisms.
  17. Our paper Not All Models Are Explained Equal: How Explainable Machine Learning Suffers from the Rashomon Effect got accepted at ECMLPKDD. See you in Torino, Italy!
  18. I have joined TU Wien's Machine Learning Research Unit in May 2023.
  19. Two accepted papers at ACL'2023! Hidden Schema Networks and A New Aligned Simple German Corpus. See you in Toronto, Canada!
  20. Our paper on Expectation-Complete Graph Representations with Homomorphisms has been accepted at ICML'23! See you in Honolulu, Hawaii!

Current Preprints

  1. Raffaele Paolino*, Sohir Maskey*, Pascal Welke, Gitta Kutyniok (2024):
    Weisfeiler and Leman Go Loopy: A New Hierarchy for Graph Representational Learning.
    Advances in Neural Information Processing Systems (NeurIPS)
    (accepted as oral presentation)

    [poster] [code] [reviews] [arXiv] [conference]

  2. Franka Bause*, Fabian Jogl*, Patrick Indri, Tamara Drucks, David Penz, Nils Morten Kriege, Thomas Gärtner, Pascal Welke, Maximilian Thiessen (2024):
    Maximally Expressive GNNs for Outerplanar Graphs.
    Transactions on Machine Learning Research (TMLR)
    (accepted for publication)

    [pdf] [code] [reviews] [journal]

  3. Fabian Jogl, Pascal Welke, Thomas Gärtner (2024):
    Is Expressivity Essential for the Predictive Performance of Graph Neural Networks?.
    Workshop on Scientific Methods for Understanding Deep Learning (SciForDL@NeurIPS)
    (accepted as poster presentation)

    [code] [reviews] [workshop]

Publications

  1. Alexander Pluska, Pascal Welke, Thomas Gärtner, Sagar Malhotra (2024):
    Logical Distillation of Graph Neural Networks.
    International Conference on Knowledge Representation and Reasoning (KR)
    (honorable mention award at the Special Track on Reasoning, Learning, and Decision Making)

    [pdf] [poster] [code] [arXiv] [conference]

  2. Fouad Alkhoury, Pascal Welke (2024):
    Splitting Stump Forests.
    International Conference on Discovery Science (DS)
    (Best Student Paper Award)

    [pdf] [slides] [code] [conference]

  3. Sebastian Müller, Vanessa Toborek, Katharina Beckh, Matthias Jakobs, Christian Bauckhage, Pascal Welke (2023):
    An Empirical Evaluation of the Rashomon Effect in Explainable Machine Learning.
    European Conference on Machine Learning and Principles and Practice of Knowledge Discovery in Databases (ECMLPKDD)

    [pdf] [code] [doi] [arXiv] [conference]

  4. Pascal Welke*, Maximilian Thiessen*, Fabian Jogl, Thomas Gärtner (2023):
    Expectation-Complete Graph Representations with Homomorphisms.
    International Conference on Machine Learning (ICML)

    [pdf] [poster] [slides] [video] [code] [reviews] [arXiv] [conference]

  5. Ramsés J. Sánchez, Lukas Conrads, Pascal Welke, Kostadin Cvejoski, César Ojeda (2023):
    Hidden Schema Networks.
    Annual Meeting of the Association for Computational Linguistics (ACL)

    [pdf] [poster] [slides] [code] [doi] [arXiv] [bibtex] [conference]

  6. Vanessa Toborek, Moritz Busch, Malte Boßert, Christian Bauckhage, Pascal Welke (2023):
    A New Aligned Simple German Corpus.
    Annual Meeting of the Association for Computational Linguistics (ACL)

    [pdf] [poster] [code] [doi] [arXiv] [bibtex] [conference]

  7. Katharina Beckh, Sebastian Müller, Matthias Jakobs, Vanessa Toborek, Hanxiao Tan, Raphael Fischer, Pascal Welke, Sebastian Houben, Laura von Rüden (2023):
    Harnessing Prior Knowledge for Explainable Machine Learning: An Overview.
    IEEE Conference on Secure and Trustworthy Machine Learning (SatML)

    [pdf] [video] [doi] [reviews] [arXiv] [bibtex] [conference]

  8. Till Hendrik Schulz, Tamás Horváth, Pascal Welke, Stefan Wrobel (2022):
    A generalized Weisfeiler-Lehman graph kernel.
    Machine Learning (111)

    [pdf] [code] [doi] [arXiv] [bibtex] [journal]

  9. Dario Antweiler, Marc Harmening, Nicole Marheineke, Andre Schmeißer, Raimund Wegener, Pascal Welke (2022):
    Machine learning framework to predict nonwoven material properties from fiber graph representations.
    Software Impacts (14)

    [pdf] [code] [reproducible run] [doi] [bibtex] [journal]

  10. Dario Antweiler, Marc Harmening, Nicole Marheineke, Andre Schmeißer, Raimund Wegener, Pascal Welke (2022):
    Graph-Based Tensile Strength Approximation of Random Nonwoven Materials by Interpretable Regression.
    Machine Learning with Applications (8)

    [pdf] [code] [reproducible run] [doi] [journal]

  11. Till Hendrik Schulz, Pascal Welke, Stefan Wrobel (2022):
    Graph Filtration Kernels.
    AAAI Conference on Artificial Intelligence (AAAI)

    [pdf] [poster] [slides] [code] [doi] [arXiv] [bibtex] [conference]

  12. Richard Palme, Pascal Welke (2022):
    Frequent Generalized Subgraph Mining via Graph Edit Distances.
    IoT Streams for Predictive Maintenance (IoTStreams@ECMLPKDD)

    [pdf] [slides] [code] [doi] [bibtex] [workshop]

  13. Janis Kalofolias, Pascal Welke, Jilles Vreeken (2021):
    SUSAN: The Structural Similarity Random Walk Kernel.
    SIAM International Conference on Data Mining (SDM)

    [pdf] [slides] [video] [code] [doi] [bibtex] [conference]

  14. Pascal Welke (2020):
    Efficient Frequent Subgraph Mining in Transactional Databases.
    International Conference on Data Science and Advanced Analytics (DSAA)

    [pdf] [slides] [video] [doi] [bibtex] [conference]

  15. Pascal Welke, Fouad Alkhoury, Christian Bauckhage, Stefan Wrobel (2020):
    Decision Snippet Features.
    International Conference on Pattern Recognition (ICPR)

    [pdf] [slides] [video] [code] [doi] [bibtex] [conference]

  16. Pascal Welke, Florian Seiffarth, Michael Kamp, Stefan Wrobel (2020):
    HOPS: Probabilistic Subtree Mining for Small and Large Graphs.
    SIGKDD Conference on Knowledge Discovery and Data Mining (KDD)

    [pdf] [slides] [video] [code] [doi] [bibtex] [conference]

  17. Alexander Mehler, Wahed Hemati, Pascal Welke, Maxim Konca, Tolga Uslu (2020):
    Multiple Texts as a Limiting Factor in Online Learning: Quantifying (Dis-)similarities of Knowledge Networks across Languages.
    Frontiers in Education | Digital Education

    [pdf] [doi] [arXiv] [bibtex] [journal]

  18. Pascal Welke, Tamás Horváth, Stefan Wrobel (2019):
    Probabilistic and Exact Frequent Subtree Mining in Graphs Beyond Forests.
    Machine Learning (108)

    [pdf] [doi] [bibtex] [journal]

  19. Pascal Welke, Tamás Horváth, Stefan Wrobel (2018):
    Probabilistic Frequent Subtrees for Efficient Graph Classification and retrieval.
    Machine Learning (107)

    [pdf] [doi] [bibtex] [journal]

  20. Till Hendrik Schulz, Tamás Horváth, Pascal Welke, Stefan Wrobel (2018):
    Mining Tree Patterns with Partially Injective Homomorphisms.
    European Conference on Machine Learning and Knowledge Discovery in Databases (ECMLPKDD)

    [pdf] [slides] [doi] [bibtex] [conference]

  21. Pascal Welke, Alexander Markowetz, Torsten Suel, Maria Christoforaki (2016):
    Three-hop Distance Estimation in Social Graphs.
    IEEE International Conference on Big Data (BigData)

    [pdf] [slides] [doi] [bibtex] [conference]

  22. Pascal Welke, Tamás Horváth, Stefan Wrobel (2016):
    Min-Hashing for Probabilistic Frequent Subtree Feature Spaces.
    International Conference on Discovery Science (DS)

    [pdf] [poster] [slides] [doi] [bibtex] [conference]

  23. Katrin Ullrich, Jennifer Mack, Pascal Welke (2016):
    Ligand Affinity Prediction with Multi-pattern Kernels.
    International Conference on Discovery Science (DS)

    [pdf] [slides] [doi] [bibtex] [conference]

  24. Pascal Welke, Ionut Andone, Konrad Blaszkiewicz, Alexander Markowetz (2016):
    Differentiating Smartphone Users by App Usage.
    International Joint Conference on Pervasive and Ubiquitous Computing (UbiComp)

    [pdf] [slides] [doi] [bibtex] [conference]

  25. Pascal Welke, Tamás Horváth, Stefan Wrobel (2015):
    Probabilistic Frequent Subtree Kernels.
    New Frontiers in Mining Complex Patterns (NFMCP@ECMLPKDD)

    [pdf] [slides] [doi] [bibtex] [workshop]

  26. Pascal Welke, Tamás Horváth, Stefan Wrobel (2014):
    On the Complexity of Frequent Subtree Mining in Very Simple Structures.
    International Conference on Inductive Logic Programming (ILP)

    [pdf] [slides] [doi] [bibtex] [conference]

  27. Anne-Kathrin Mahlein, Till Rumpf, Pascal Welke, Heinz-Wilhelm Dehne, Ulrike Steiner, Erich-Christian Oerke (2013):
    Development of Spectral Indices for Detecting and Identifying Plant Diseases.
    Remote Sensing of Environment (128)

    [doi] [journal]

Books

  1. Michael Kamp et al. (2021):
    Machine Learning and Principles and Practice of Knowledge Discovery in Databases - International Workshops of ECML PKDD 2021, Virtual Event, September 13-17, 2021, Proceedings, Part I.

    [doi] [bibtex] [workshop proceedings]

  2. Michael Kamp et al. (2021):
    Machine Learning and Principles and Practice of Knowledge Discovery in Databases - International Workshops of ECML PKDD 2021, Virtual Event, September 13-17, 2021, Proceedings, Part II.

    [doi] [bibtex] [workshop proceedings]

  3. Daniel Trabold, Pascal Welke, Nico Piatkowski (2020):
    Proceedings of the Conference "Lernen, Wissen, Daten, Analysen", Online, September 9-11, 2020.

    [bibtex] [proceedings]

  4. Pascal Welke (2019):
    Efficient Frequent Subtree Mining Beyond Forests.
    Dissertations in Artificial Intelligence (348)

    [pdf] [slides] [code] [bibtex] [book]

Non-Archival Venues

  1. Raffaele Paolino*, Sohir Maskey*, Pascal Welke, Gitta Kutyniok (2024):
    Weisfeiler and Leman Go Loopy: A New Hierarchy for Graph Representational Learning.
    Bridging the Gap Between Practice and Theory in Deep Learning (BGPT@ICLR)

    [pdf] [poster] [code] [reviews] [arXiv] [workshop]

  2. Alexander Pluska, Pascal Welke, Thomas Gärtner, Sagar Malhotra (2024):
    Logical Distillation of Graph Neural Networks.
    Mechanistic Interpretability Workshop (MI@ICML)

    [pdf] [poster] [code] [arXiv] [workshop]

  3. Veronica Lachi*, Alice Moallemy-Oureh*, Andreas Roth*, Pascal Welke* (2023):
    Graph Pooling Provably Improves Expressivity.
    New Frontiers in Graph Learning (GLFrontiers@NeurIPS)

    [pdf] [poster] [reviews] [workshop]

  4. Franka Bause*, Fabian Jogl*, Patrick Indri, Tamara Drucks, David Penz, Nils Morten Kriege, Thomas Gärtner, Pascal Welke, Maximilian Thiessen (2023):
    Maximally Expressive GNNs for Outerplanar Graphs.
    New Frontiers in Graph Learning (GLFrontiers@NeurIPS)

    [pdf] [poster] [code] [reviews] [workshop]

  5. Franka Bause*, Fabian Jogl*, Pascal Welke, Maximilian Thiessen (2023):
    Maximally Expressive GNNs for Outerplanar Graphs.
    Learning on Graphs Conference (LoG)
    (Extended Abstract)

    [pdf] [poster] [code] [reviews] [conference]

  6. Andrei Dragos Brasoveanu, Fabian Jogl, Pascal Welke, Maximilian Thiessen (2023):
    Extending Graph Neural Networks with Global Features.
    Learning on Graphs Conference (LoG)
    (Extended Abstract)

    [pdf] [poster] [code] [reviews] [conference]

  7. Maximilian Thiessen*, Pascal Welke*, Thomas Gärtner (2022):
    Expectation Complete Graph Representations using Graph Homomorphisms.
    New Frontiers in Graph Learning Workshop (GLFrontiers@NeurIPS)

    [pdf] [poster] [code] [reviews] [workshop]

  8. Pascal Welke*, Maximilian Thiessen*, Thomas Gärtner (2022):
    Expectation Complete Graph Representations using Graph Homomorphisms.
    Learning on Graphs Conference (LoG)

    [pdf] [poster] [code] [reviews] [conference]

  9. Dario Antweiler, Pascal Welke (2020):
    Temporal Graph Analysis for Outbreak Pattern Detection in COVID-19 Contact Tracing Networks.
    Machine Learning in Public Health Workshop (MLPH@NeurIPS)

    [pdf] [slides] [workshop]

  10. Till Hendrik Schulz, Pascal Welke (2018):
    On the Necessity of Graph Kernel Baselines.
    Graph Embedding and Mining Workshop, (GEM@ECMLPKDD)

    [pdf] [poster] [workshop]

  11. Pascal Welke (2017):
    Simple Necessary Conditions for the Existence of a Hamiltonian Path with Applications to Cactus Graphs.
    Computer Science Conference for University of Bonn Students (CSCUBS)

    [pdf] [arXiv] [bibtex] [workshop]

Lecture Notes and Coding Nuggets

  1. Pascal Welke, Christian Bauckhage (2021):
    Linear Programming for Robust Regression.
    ML2R Coding Nugget

  2. Christian Bauckhage, Pascal Welke (2021):
    Sorting as Linear Programming.
    ML2R Coding Nugget

  3. Christian Bauckhage, Pascal Welke (2021):
    Sorting as Quadratic Unconstrained Binary Optimization Problem.
    ML2R Coding Nugget

  4. Christian Bauckhage, Pascal Welke (2021):
    Centering Data- and Kernel Matrices.
    ML2R Theory Nugget

  5. Pascal Welke, Till Hendrik Schulz, Christian Bauckhage (2021):
    Computational Complexity of Max-Sum Diversification.
    ML2R Theory Nugget

  6. Christian Bauckhage, Pascal Welke (2021):
    Solving Least Squares Gradient Flows.
    ML2R Coding Nugget

  7. Pascal Welke, Christian Bauckhage (2020):
    Solving Linear Programming Problems.
    ML2R Coding Nugget

Community Activities

  1. I am a regular program committee member of conferences, e.g.,
  2. I peer-review for workshops, e.g.,
  3. We have organized MLG@ECMLPKDD 2023, the 20th Workshop on Mining and Learning with Graphs. Join us for interesting discussions about graphs in Torino, Italy!
  4. I was awarded as one of the 20 top LOG'22 reviewers!
  5. We have organized MLG@ECMLPKDD 2022, the 18th Workshop on Mining and Learning with Graphs.
  6. I was program chair of the KDML track at LWDA 2022.
  7. We have organized GEM'21, the third Workshop on Graph Embedding and Mining, collocated with ECMLPKDD'21! It has been a pleasure! Here are the proceedings.
  8. I have co-organized GEM'20, the Workshop on Graph Embedding and Mining, collocated with ECMLPKDD'20.
  9. I was program chair (with Nico Piatkowski) of the KDML track at LWDA 2020. It has been a pleasure. Here are the proceedings.
  10. Member of the program committee of GEM'19, the Workshop on Graph Embedding and Mining, collocated with ECMLPKDD'19.
  11. Reviews for several journals, conferences, and academic funding programs, e.g. Machine Learning, Data Mining and Knowledge Discovery, AMAI, ACM SIGKDD 2016, KI-Starter NRW.

This page tries to be minimalistic in layout, bandwith, and used tools. It is hosted on github pages, using neat.css stylesheets, and bibtexparser to generate the lists of references.