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.

relNet Opening Workshop

The project partners of LS8, Dortmund and CERES, Bochum host an opening workshop for their new joint project relNet on "Modelling Topics and Structures in Religious Online Communication" in Bochum from May 23-24. The goal of this project is to apply methods of data analytics, network analysis and text mining to analyse how digital communication has changed religious communities and the social roles within these communities.

On to days we want to present the project, listen to talks by our invited guests and discuss the potentials of joint research in computer science and the social sciences, in this case religious studies. Click below for the full program.


(more...  )

Springer Edited Volume `Compuational Sustainability' published

Katharina Morik and Kristian Kersting together wit Jörg Lässig from the University of Applied Sciences Zittau/Görlitz have published an edited volume on Computational Sustainability. Computational Sustainability is a broad field that attempts to optimize societal, economic, and environmental resources using methods from computer science, mathematics and related fields:
Springerl Jörg Lässig, Kristian Kersting, Katharina Morik, Computational Sustainability. Studies in Computational Intelligence, Volume 645 2016, Springer, ISBN: 978-3-319-31856-1, 2016.
(more...  )

Best Paper Award of the AAAI Conference on Artificial Intelligence and Interactive Digital Entertainment (AIIDE) 2015

The joint work "Predicting Purchase Decisions in Free To Play Mobile Games" of Kristian Kersting with colleagues from Wooga, goedle.io, Aalborg University, and the Fraunhofer IAIS received the Best Paper Award of the AAAI Conference on Artificial Intelligence and Interactive Digital Entertainment (AIIDE) 2015.

LS8 is an international collaborator of CompSusNet

LS8 is an international collaborator of CompSusNet. CompSustNet is a research network sponsored by the National Science Foundation through an Expeditions in Computing award. Twelve U.S. academic institutions led by Cornell University, along with many national and international collaborators, are exploring new research directions in computational sustainability. (more...  )

Morgan and Claypool Book on Statistical Relational AI

Together with colleagues from UBC, KU Leuven, and U. Indiana, Kristian Kersting published a book on Statistical Relational AI. This is the study and design of intelligent agents that act in worlds composed of individuals (objects, things), where there can be complex relations among the individuals, where the agents can be uncertain about what properties individuals have, what relations are true, what individuals exist, whether different terms denote the same individual, and the dynamics of the world:

Morgan and Claypool Luc De Raedt, Kristian Kersting, Sriraam Natarajan, David Poole, Statistical Relational Artificial Intelligence: Logic, Probability, and Computation. Morgan and Claypool Publishers,Synthesis Lectures on Artificial Intelligence and Machine Learning, ISBN: 9781627058414, 2016.
(more...  )

VaVeL Project

Urban environments are flooded with data from fixed or mobile sensors that are gathering data. If these data were used successfully, European citizens could benefit in various areas like public transport or crime prevention. However, urban data is heterogenous, noisy and unlabeled, since the usability of the data is low. The VaVeL project aims towards using these data in application for increasing the living conditions in urban areas. The goal of the project is developing a general framework for managing and mining heterogenous urban data streams.

As part of the project, the functionality of current stream frameworks shall be fit on data streams from urban sensors. The access to urban data streams is not an easy task; In this project, a set of black boxes will be implemented to give easier access to the data and analysis procedures. Big data companies shall get an access to the gathered knowledge so that actual problems of an urban environment can be tackled.

(more...  )

Call for Papers - Data Mining for Smart Cities

There is a Call for Papers for the journal Data Mining for Smart Cities. They are looking, for example, for the following topics:

  • Real-time nowcasting and prediction of events
  • Interactive exploration of city data
  • Feature extraction and deep learning from urban data

Submission is due July 4, 2016, 23:59.

(more...  )

Katharina Morik as new member of acatech - National Acadamy of Science and Engineering

Katharina Morik

The National Acadamy of Science and Engineering advises society and governments in all questions regarding the future of technology. Acatech is one of the most important academies for novel technology research. Additionally, acatech provides a platform for transfer of concepts to applications and enables the dialogue between science and industry. The members work together with external researchers in interdisciplinary projects to ensure the practiability of recent trends. Internationally oriented, acatech wants to provide solutions for global problems and new perspectives for technological value added in Germany.

By the appointment of Katharina Morik as member of acatech, the acadamy recognizes her research profile, her achievements as speaker of the collaborative research center SFB 876, her international reputation and innovative research in machine learning.

Christian Bockermann Defends his Dissertation at LS8

Christian Bockermanns Disputation

Christian Bockermann has successfully defended his dissertation with the title “Mining Big Data Streams for Multiple Concepts”. His thesis was supervised by Katharina Morik. Summary of the thesis:

Modelling streaming data applications in near real-time is motivated by today’s growing demand for in-time data analysis. The thesis reviews the Lambda architecture and state of the art frameworks for data streams and introduces a middle-layer easing the definition of streaming applications in a platform independent way. This enabling technique is demonstrated in two Big Data applications, namely the inline processing and analysis of data in Cherenkov astronomy and the near real-time extraction of viewership statistics in the context of an IP-TV platform.

(more...  )

Fabian Hadiji Defends his Dissertation at LS8

Fabian Hadiji successfully defended his dissertation under the title "Graphical Models Beyond Standard Settings: Lifted Decimation, Labeling, and Counting". His thesis was supervised by Professor Kristian Kersting.

He summarises his thesis in the following abstract:

With increasing complexity and growing problem sizes in AI and Machine Learning, inference and learning are still major issues in Probabilistic Graphical Models (PGMs). On the other hand, many problems are specified in such a way that symmetries arise from the underlying model structure. Exploiting these symmetries during inference, which is referred to as "lifted inference", has lead to significant efficiency gains. This thesis provides several enhanced versions of known algorithms that show to be liftable too and thereby applies lifting in "non-standard" settings. By doing so, the understanding of the applicability of lifted inference and lifting in general is extended. Among various other experiments, it is shown how lifted inference in combination with an innovative Web-based data harvesting pipeline is used to label author-paper-pairs with geographic information in online bibliographies. This results is a large-scale transnational bibliography containing affiliation information over time for roughly one million authors. Analyzing this dataset reveals the importance of understanding count data. Although counting is done literally everywhere, mainstream PGMs have widely been neglecting count data. In the case where the ranges of the random variables are defined over the natural numbers, crude approximations to the true distribution are often made by discretization or a Gaussian assumption. To handle count data, Poisson Dependency Networks (PDNs) are introduced which presents a new class of non-standard PGMs naturally handling count data.

If you are interested in Fabian's past and future work, also see his personal homepage http://hadiji.com/.

Second On the Record at the Signal Iduna Park

Being at the famous stadion of the Dortmund football team BVB 09, we could not resist to pretend giving a press conference. Actually, the conference that we visited was on economic journalism in the digital age. http://www.wipojo.de/ontherecord/

Best Paper Presentation Award of the "New Challenges in Neural Computation" Workshop 2015

The joint work "Archetypal Analysis as an Autoencoder" Of Kristian Kersting with colleagues from the University of Bonn and the Twenty Billion Neurons GmbH received the Best Presentation Award of the "Challenges in Neural Computation" (NC^2) Workshop of the GI-Fachgruppe Neuronale Netze and the German Neural Networks Society in connection to GCPR 2015, Aachen.

Successful Final Review of the EU Project INSIGHT in Luxemburg

The goal of the INSIGHT project was to radically advance our ability of coping with emergency situations in smart cities. INSIGHT stands for Intelligent Synthesis and Real-time Response using massive Streaming of Heterogeneous Data and the developed technologies for data stream mining put new capabilities in the hands of disaster planners and city personell to improve emergency planning and response.

(more...  )

LS8 in Banff, Canada

From July 24th to July 26th, the LS8 gave two talks at the Workshop "Advances in interactive Knowledge Discovery and Data Mining in Complex and Big Data Sets" in Banff, Canada.

Professor Katharina Morik spoke about "Big Data and Small Devices": Analyzing data on small devices confronts us with new challanges with regard to runtime, memory consumption and energy consumption. Her talk investigates the use of graphical models for data mining in resource-restricted environments and presents results from the research project SFB876. 

Furthermore,Sibylle Hess presented results of her diploma thesis: "Investigation of Code Tables to Compress and Describe Underlying Characteristics of Binary Databases". She connects traditional methods of frequent pattern mining with the Minimum-Description-Length principle and matrix factorization, combining these techniques into new algorithms for frequent pattern mining based on numerical optimization.

2nd Workshop on Mining Urban Data held in conjunction with ICML

We co-organize this years 2nd Workshop on Mining Urban Data. The workshop takes place July 11th at ICML Lille. Please see the proceedings at http://ceur-ws.org/Vol-1392/ This year we welcome three invited speakers: * Dr. Eleni Pratsini - "Using Big Mobile Data to Analyze Social Events in Cities" * Prof. Kristian Kersting - "Poisson Dependency Networks: Gradient Boosted Models for Multivariate Count Data" * Prof. Sharad Mehrotra - Towards `on the fly' data cleaning (more...  )

Stellen für studentische Hilfskräfte ab sofort

An der TU Dortmund, Fakultät für Informatik am Lehrstuhl VIII, sind ab sofort Stellen für Studentische Hilfskräfte zu besetzen.

(more...  )

Summer school

Summer School 2015

The next summer school will be hosted at the faculty of sciences of the university of Porto from 2nd to 5th of September and is collocated with ECMLPKDD 2015. It will be organize by LIAAD-INESC TEC and TU Dortmund.

For the summer school, world leading researchers in machine learning and data mining will give lectures on recent techniques for example dealing with huge amounts of data or spatio-temporal streaming data.

(more...  )

SHE - Sie hat's erfunden

Bei den Unternhemenstagen , die vom 26.01.2015 bis zum 09.02.2015 stattfanden, waren Frauen im Berufsleben einer der Schwerpunkte der Veranstaltungsreihe. Um die Innovationen von klein und mittelständigen Unternehmen zu fördern, muss die Wahrnehmung von Frauen als Erfinderinnen gefördert werden. Professorin Katharina Morik nahm an einer Gesprächsrunde teil, die unter anderem über die Vereinbarkeit von Familie und Berufsleben und die mangelhafte Wahrnehmung von Frauen als Erfinderinnen diskutierte.

(more...  )

Datenanalysten werden in der Wirtschaft gesucht

Datenanalyse ist in der Wirtschaft die gesuchte Kompetenz — die Lehre des LS 8 liefert den Studierenden auch im Sommersemester 2015 wieder das Wissen dazu! (more...  )

RapidMiner Academics - Free Access to RapidMiner Studio for Students

RapidMiner CEO and founder Ingo Mierswa presents RapidMiner Academia: A program that grants students free access to commercial versions of RapidMiner Studio.

The RapidMiner project began in 2001, at that point still called YALE, at the LS8 here at TU Dortmund. Today it is one of the most popular software environments for predictive Data Analysis and Data Mining.

(more...  )

Celebratory Colloquium at the Faculty for Computer Science

Katharina Morik

Ein besonderes Spektrum an Vorträgen fand am Jahresende zum 60. Geburtstag von Katharina Morik statt. Gemeinsam war den drei Hauptrednern, dass sie im Bereich Maschinelles Lernen bzw. Data Mining international höchst renommiert sind und bei Katharina Morik an der (Technischen) Universität Dortmund promovierten. Völlig verschieden ihre Tätigkeitsfelder.

Inhaltliche Gemeinsamkeiten der Redner mit der Jubilarin wurden in der kurzen Einführung deutlich, in der Katharina Morik ihre Forschungsziele zusammenfasste, die sie an der TU Dortmund verfolgt: situierte Systeme, die durch Lernfähigkeit Sensorik, Kommunikation und Handlung verbinden. Anfang der 90er Jahre entstanden Arbeiten zur Robotik: realzeitlich wurden in verteilten, heterogenen Datenströmen Muster entdeckt, die zur Handlungsplanung eingesetzt wurden. Der SFB 876 (Informatik), dessen zweite Phase gerade bewilligt wurde, kann mit seiner Verbindung von Datenanalyse und Cyber Physical Systems in der Leitlinie lernfähiger, situierter Systeme gesehen werden.

Die Arbeiten zu sehr großen Datenmengen, die Katharina Morik in 12 Jahren im SFB 475 (Statistik) zusammen mit Claus Weihs durchgeführt hat, wurden von der Sprecherin dieses Sonderforschungsbereichs, Ursula Gather, in einer kurzen Ansprache gewürdigt.

Ganz unterschiedliche Herangehensweisen, maschinelles Lernen erfolgreich zu erforschen und anzuwenden, wurden durch die Hauptvorträge der drei herausragenden Wissenschaftler deutlich.

Stefan Wrobel Thorsten Joachims Ingo Mierswa

Unsere Studierenden mag es freuen, wenn sie einige Beispiele sehen, wozu das Studium an der TU Dortmund befähigt: Forschungsdirektor, Professor, CEO einer Firma – auf der Grundlage herausragender Forschung zu maschinellem Lernen, Data Mining, Big Data Analytics lässt sich einiges machen!


(more...  )
Show news archive