Hauptnavigation

News from the Artificial Intelligence Group

The chair of artificial intelligence deals with the wide field of machine learning. In particular the chair concentrates on the development and implementation of learning algorithms that solve challenging problems.






Lukas Pfahler defends his dissertation at LS8

Jahreskonferenz der Plattform Lernende Systeme in Berlin

Lukas Pfahler has defended his dissertation Some Representation Learning Tasks and the Inspection of Their Models with distinction (summa cum laude). In his work, he focuses on representation learning with unsupervised methods. For instance, he has investigated the use of embedding learning with graph convolutional neural networks for the search and retrieval of related mathematical expressions. Furthermore, he has worked on novel methods for model inspection to increase trust in decisions.

The members of the PhD committee were Prof. Dr. Katharina Morik (supervisor and first reviewer), Prof. Dr. Andreas Hotho (second reviewer, University of Würzburg), Prof. Dr. Jakob Rehof (chair) and Priv.-Doz. Dr. habil Frank Weichert (faculty representative). Lukas Pfahler is a research associate at LS8 and member of the Collaborative Research Center 876 (project A1).

Mirko Bunse defends his dissertation at LS8

Jahreskonferenz der Plattform Lernende Systeme in Berlin

Mirko Bunse has defended his dissertation Machine Learning for Acquiring Knowledge in Astro-Particle Physics with great praise (magna cum laude). In his work, he studied the manifold applications of Machine Learning algorithms in Astroparticle Physics. In particular, he focuses on the smart and resource-aware control of simulations through active class selection and the domain-specific aggregation of predictions in terms of quantification and unfolding.

The members of the PhD committee were Prof. Dr. Katharina Morik (supervisor and first reviewer), Dr. Fabrizio Sebastiani (second reviewer, Consiglio Nazionale delle Ricerche, Pisa), Prof. Dr. Johannes Fischer (chair) and Jun.-Prof. Dr. Thomas Liebig (faculty representative). Mirko Bunse is a research associate at LS8, member of the Collaborative Research Center 876 (project C3) and Coordinator of the application field astroparticle physics at the Lamarr Institute for Machine Learning and Artificial Intelligence (former Competence Center ML2R).

Katharina Morik organizes "Smart City" Session at the Third Trilateral Symposium on Artificial Intelligence of Japan, Germany and France in Tokyo

The 3rd trilateral AI symposium of Japan, Germany and France took place the 27th of October 2022 in Tokyo. Katharina Morik organized the session “Smart Cities” in which 2 speakers of each country presented their work. She presented the EU projects INSIGHT and VAVEL (coordinator. Dimitrios Gunopoulos), in which she participated together with Thomas Liebig. The concluding discussion stressed as most important the acquisition of mobility data and the interaction of all the stakeholders. How should mobility companies, governmental institutions, IT companies and the users cooperate?

Best Student Paper Award: Faster Silhouette Clustering

At the SISAP 2022 conference at the University of Bologna, Lars Lenssen won the "best student paper" award for the contribution "Lars Lenssen, Erich Schubert. Clustering by Direct Optimization of the Medoid Silhouette. In: Similarity Search and Applications. SISAP 2022. https://doi.org/10.1007/978-3-031-17849-8_15".

The publisher Springer donates a monetary prize for the awards, and the best contributions are invited to submit an extended version to a special issue of the A* journal "Information Systems".

In this paper, we introduce a new clustering method that directly optimizes the Medoid Silhouette, a variant of the popular Silhouette measure of clustering quality. As the new variant is O(k²) times faster than previous approaches, we can cluster data sets larger by orders of magnitude, where large values of k are desirable. The implementation is available in the Rust "kmedoids" crate and the Python module "kmedoids", the code is open source on Github.

The group is successful for the second time: In 2020, Erik Thordsen won the award with the contribution "Erik Thordsen, Erich Schubert. ABID: Angle Based Intrinsic Dimensionality. In: Similarity Search and Applications. SISAP 2020. https://doi.org/10.1007/978-3-030-60936-8_17".

This paper introduced a new angle-based estimator of the intrinsic dimensionality – a measure of local data complexity – traditionally estimated solely from distances.

Sebastian Buschjäger defends his dissertation at LS8

Jahreskonferenz der Plattform Lernende Systeme in Berlin

Sebastian Buschjäger has defended his dissertation Ensemble Learning with Discrete Classifiers on Small Devices at the Chair of Artificial Intelligence with distinction (summa cum laude). He conducted research on the topic of resource-aware machine learning in the context of the Collaborative Research Center 876, Project A1. He researched ensemble methods in the context of embedded systems. This included training as well as deploying decision forests on small devices.


The members of the PhD committee were Prof. Dr. Katharina Morik (supervisor and first reviewer), Prof. Johannes Fürnkranz (second reviewer, University of Linz), Prof. Dr. Jian-Jia Chen (chair) and Prof. Dr. Jens Teubner (faculty representative). Sebastian Buschjäger is a research associate at LS8 and a member of the Collaborative Research Center 876 (Project A1).

Vordenker Forum 2022

The technology pioneer for autonomous driving, Prof. Sebastian Thrun, was honored as "Vordenker 2022" at Goethe University Frankfurt on September 15. The former Google vice-director and Stanford professor founded the online learning platform Udacity and is now dedicated to autonomous flying. In his speech, he recognized Katharina Morik as a leading pioneer in the field of artificial learning. Prof. Thrun said she "was already the goddess of artificial learning back then. The very first one who did it in Germany and is still quite a leader today." In the panel, Prof. Morik explained how artificial intelligence can help make work more productive, safe and environmentally friendly, creating capacity for social good. The director of the Lamarr Institute for Machine Learning and Artificial Intelligence also presented Lamarr's research on intelligible communication of artificial intelligence in the form of so-called care labels.

A recording of the event is available online: 

Student assistant wanted in the field of Federated Learning (SHK / WHF) - the job has been taken

The Chair VIII of the Faculty of Computer Science has an immediate vacancy for a student assistant (SHK / WHF) in the field of Federated Learning. The number of hours can be discussed individually. The offer is aimed at students of computer science who have completed their studies with very good results.

You can find more information about the positions and your application here

Student assistants (SHK / WHF) wanted - the job has been taken

The Chair VIII of the Faculty of Computer Science has immediate vacancies for student assistants (SHK / WHF). The number of hours can be discussed individually. The offer is aimed at students of computer science who have completed their studies with very good results.  

You can find more information about the positions and your application here

Best Paper Award at the ICDM PhD Forum

We are pleased to announce that Pierre Haritz, Helena Kotthaus, Thomas Liebig and Lukas Pfahler have received the "Best Paper Award" for the paper "Self-Supervised Source Code Annotation from Related Research Papers" at the IEEE ICDM PhD Forum 2021.

To increase the understanding and reusability of third-party source code, the paper proposes a prototype tool based on BERT models. The underlying neural network learns common structures between scientific publications and their implementations based on variables occurring in the text and source code, and will be used to annotate scientific code with information from the respective publication.

(more information  )

6Gem Project

6Gem LogoThe 6GEM consortium combines scientific excellence and mobile communications expertise at network, material, component/microchip, and module-level in North Rhine-Westphalia. A holistic approach is pursued, from production to logistics to people with their needs for self-determination, privacy, and security in times of climate change.

Based on previous contributions in the SFB 876, the LS8 project team will explore novel, real-time capable 6G network technologies and innovative 6G application fields. Among other things, the results will flow into the standardization of open 6G networks, open-source projects for software-defined networks, and patents.

Prof. Christian Wietfeld from the Department of Communication Networks is the spokesperson for the TU Dortmund in the 6GEM project. Also involved from the Faculty of Electrical Engineering and Information Technology are Embedded Systems, High-Frequency Technology, and Energy Efficiency. From the Faculty of Computer Science, the areas of Design Automation for Embedded Systems and Smart City Science are also involved. From the Faculty of Mechanical Engineering, the area of Materials Handling and Warehousing.

(more information  )

Volkswagen Foundation Report on reflecive AI has been published

The report of the project on reflective AI, funded by the Volkswagen Stiftung, is published. It is about the user’s awareness of the implications of AI systems.


Ensuring a safe and responsible use of AI cannot be solved alone through technological innovation and regulation, in spite of their importance. Many of the problems encountered in the use of AI systems stem from the lack of personal and societal experience with AI. They mirror not only the biases and inequalities reflected in the data and AI algorithms but also those from the organizational and societal contexts in which AI is used and designed. 


View report.

Mirko Bunse and Lukas Heppe among the Winners of the Ariel Machine Learning Data Challenge

Scientists Mirko Bunse (Collaborative Research Center/SFB 876) and Lukas Heppe (ML2R) took second place in the Ariel Machine Learning Data Challenge at the ECML PKDD 2021 conference. Together, they developed a multi-level Deep Learning method for analyzing noisy time series data. Using data preprocessing, they bundled information from the data set, including noise properties. This bundling of information allowed for a training of neural networks that is efficient enough to create an ensemble of 45 individual networks. The developed approach achieved an average prediction error of only three percent.

More informationen to Ariel Machine Learning Data Challenge

Stefanie Jegelka from MIT gives a talk on "Learning in Graph Neural Networks"

Event date: July 15 2021 16:15

Learning in Graph Neural Networks

Abstract - Graph Neural Networks (GNNs) have become a popular tool for learning representations of graph-structured inputs, with applications in computational chemistry, recommendation, pharmacy, reasoning, and many other areas. In this talk, I will show some recent results on learning with message-passing GNNs. In particular, GNNs possess important invariances and inductive biases that affect learning and generalization. We relate these properties and the choice of the “aggregation function” to predictions within and outside the training distribution.

This talk is based on joint work with Keyulu Xu, Jingling Li, Mozhi Zhang, Simon S. Du, Ken-ichi Kawarabayashi, Vikas Garg and Tommi Jaakkola.

 

Short bio - Stefanie Jegelka is an Associate Professor in the Department of EECS at MIT. She is a member of the Computer Science and AI Lab (CSAIL), the Center for Statistics, and an affiliate of IDSS and the ORC. Before joining MIT, she was a postdoctoral researcher at UC Berkeley, and obtained her PhD from ETH Zurich and the Max Planck Institute for Intelligent Systems. Stefanie has received a Sloan Research Fellowship, an NSF CAREER Award, a DARPA Young Faculty Award, a Google research award, a Two Sigma faculty research award, the German Pattern Recognition Award and a Best Paper Award at the International Conference for Machine Learning (ICML). Her research interests span the theory and practice of algorithmic machine learning.

 

 

Digitaltag 2021: Competence Center ML2R offers Hands-On Workshop on Machine Learning

As part of the Digitaltag (Digital Day) 2021, the Competence Center Machine Learning Rhine-Ruhr (ML2R) is hosting a joint virtual hands-on workshop with the software manufacturer RapidMiner. The German-language event under the motto "Machine Learning: An Introduction to the Key Technology of Artificial Intelligence" on 18 June offers participants of all backgrounds exciting insights into the basics of Machine Learning (ML) as well as illustrative application examples. Using the graphical software RapidMiner, participants will also learn about a typical ML workflow. Based on a concrete application task, the steps of data preparation, model building, training, prediction, and validation will be explained and exemplarily executed in the software RapidMiner Studio under the guidance of an AI trainer.

Registration for the event is free of charge. To register, please write an email to ann-kathrin.oster@tu-dortmund.de. You will then receive the access data for the event as well as instructions on the free of cost installation of the "RapidMiner Studio" software, which is essential for the practical part of the workshop.

Research and Science Officer (m/f/d) wanted - Ref.-No. 083/21e

This position should be filled in the Faculty of Computer Science in Collaborative Research Center 876 (SFB 876) as soon as possible until December 31, 2022. According to the public tariff regulations, the salary is based on the tariff group E13 TV-L.

For details: https://karriere.tu-dortmund.de/job/view/810/research-and-science-officer-m-f-d-ref-no-083-21e?page_lang=en

Student assistants (SHK / WHF) wanted - the job has been taken

The Chair VIII of the Faculty of Computer Science has an immediate vacancy for an student assistant (SHK / WHF). The number of hours can be discussed individually. The offer is aimed at students of computer science who have completed their studies with very good results. Specifically, the job is about the implementation and further development of existing procedures as well as their evaluation on small devices.

You can find more information about the position and your application here

„She transforms IT“ strengthens womens‘ role in IT

The initiative "She transforms IT" is dedicated to empowering women and girls in IT and aims to increase their participation in digitization. To this end, the initiative works, among other things, to promote digital competence among girls, to make women in IT more visible and offer them research and teaching offers which are diverse and offer extensive networking opportunities.

Prof. Dr. Katharina Morik, Professor of Artificial Intelligence and spokesperson for the Competence Center ML2R and Collaborative Research Center 876, was among the first 50 signatories of the initiative, which was presented at the Digital Summit 2020.

Competence Center ML2R starts Blog on Machine Learning and Artificial Intelligence

The Competence Center Machine Learning Rhine-Ruhr (ML2R) has launched its new blog: https://machinelearning-blog.de. In the categories Application, Research and Foundations, researchers of the Competence Center and renowned guest authors provide exciting insights into scientific results, interdisciplinary projects and industry-related findings surrounding Machine Learning (ML) and Artificial Intelligence (AI). The Competence Center ML2R brings forward-looking technologies and research results to companies and society.

Seven articles already await readers: a four-part series on ML-Basics as well as one article each within the sections Application, Research and Foundations. The authors illustrate why AI must be explainable, how obscured satellite images can be recovered using Machine Learning and show methods for the automated assignment of keywords for short texts.

Student assistants (SHK / WHF) wanted - the job has been taken

At the Faculty of Computer Science, Chair VIII, there are vacancies for assistants (SHK / WHF) to be immediately taken. The number of hours can be discussed individually. The offer is aimed at students of the TU Dortmund and FH Dortmund who have experience in researching facts and also have web programming skills. The call for applications is explicitly not aimed exclusively at students from the Department of Computer Science, but is also open to students with other study focuses who have the relevant qualifications.

More information about the position and your application can be found here

LS8 secretariat closed until 02.01.2022

The LS8 office will not be staffed between 24.12.2021 and 02.01.2022. During this time the TU Dortmund is completely closed.

With good wishes for a peaceful holiday season and a confident, healthy start into the year 2022! Your LS8-Team

Show news archive
Statuen der Osterinsel