• Zielgruppen
  • Suche
 

Courses

You can find the timetables of our study program here. 

Courses at ikg

Research Project


>> Find course on Stud.IP <<

Goal of the project

In the research project students learn to apply their theoretical knowledge to specific problems. In the project, first a research plan has to be set up, which is discussed and refined with the supervisor. Then, the student will conduct the project very independently. Students meet regularly (4-5 times) with the supervisors to discuss their progress. The research will be summarized in a short report (4-5 pages).

Topics

Exemplary topics are:

  • Implementation of an approach from the literature
  • Conducting measurements and in-depth analysis of the data
  • Simulation study

Projects

You can find a list of projects here. There is also the possibility to contact us if you have a specific idea for a project.

Big Geospatial Data

Dozenten: Prof. Martin Werner, Fabian Bock

>> Zur Veranstaltung im Stud.IP <<

Die Studierenden bekommen einen Überblick über Methoden und Infrastrukturen zum parallelen Rechnen mit sehr großen Datenmengen und über Methoden der parallelen Verarbeitung von Geoinformationen. Sie sind am Ende des Moduls in der Lage, geeignete Frameworks und Ansätze zur Umsetzung von Projekten zu bewerten und selbständig einzusetzen.

Zunächst werden unterschiedliche Methoden zur parallelen Berechnung diskutiert (z.B. Prozesse, Threads, Semaphoren, OpenMP, CUDA/GPGPU, MPI, VGAS-Systeme, Hadoop, MapReduce, NoSQL, Spark). Im Anschluss werden gängige Verfahren zum Umgang mit Ortsdaten erarbeitet. Dabei werden zusammenfassende Berechnungen (z.B. Mittelwerte, Location Entropy, KDE, Rasterisierung, Hotspot-Erkennung), Daten-Lokalität (z.B. Space Filling Curves und Geohash, Space-Time-Cubes, Clustering), Verarbeitung von Navigations- und Bewegungsdaten und weitere Themen an Beispielen diskutiert.

Praxisprojekt Topographie

Dozenten: Frank Thiemann, Malte Jan Schulze

>> Zur Veranstaltung im Stud.IP <<

Spatial and Spatiotemporal Statistics and Big Data

Dozenten: Prof. Dr. Phlipp Otto

Students are encouraged to critically analyse the performance of classical, statistical approaches for modelling spatial data under the presence of big and/or high-dimensional data. In this regard, students learn key concept of spatial and spatiotemporal statistics. Furthermore, an own simulation study is performed and described in a seminar paper.  

In a first part, important concepts of spatial and spatiotemporal statistics are introduced/repeated. In particular, the focus is on kriging and modelling spatial and spatiotemporal dependence by linear approaches, like autoregressive models. Further, we examine these approaches under the presence are large/big spatial data (incl. data streams) and discuss different approaches for reducing complexity and dimensionality. In the second half of the semester, all students work on a seminar paper assessing the performance of one of the introduced concepts for data with increasing size/complexity/dimensionality. Generally, these papers should include a small simulation study. There are individual obligatory meetings during the second half. The results of the seminar papers are presented in a colloquium.

 

 

Research Project


>> Find course on Stud.IP <<

Goal of the project

In the research project students learn to apply their theoretical knowledge to specific problems. In the project, first a research plan has to be set up, which is discussed and refined with the supervisor. Then, the student will conduct the project very independently. Students meet regularly (4-5 times) with the supervisors to discuss their progress. The research will be summarized in a short report (4-5 pages).

Topics

Exemplary topics are:

  • Implementation of an approach from the literature
  • Conducting measurements and in-depth analysis of the data
  • Simulation study

Projects

You can find a list of projects here. There is also the possibility to contact us if you have a specific idea for a project.

Big Geospatial Data

Dozenten: Prof. Martin Werner, Fabian Bock

>> Zur Veranstaltung im Stud.IP <<

Die Studierenden bekommen einen Überblick über Methoden und Infrastrukturen zum parallelen Rechnen mit sehr großen Datenmengen und über Methoden der parallelen Verarbeitung von Geoinformationen. Sie sind am Ende des Moduls in der Lage, geeignete Frameworks und Ansätze zur Umsetzung von Projekten zu bewerten und selbständig einzusetzen.

Zunächst werden unterschiedliche Methoden zur parallelen Berechnung diskutiert (z.B. Prozesse, Threads, Semaphoren, OpenMP, CUDA/GPGPU, MPI, VGAS-Systeme, Hadoop, MapReduce, NoSQL, Spark). Im Anschluss werden gängige Verfahren zum Umgang mit Ortsdaten erarbeitet. Dabei werden zusammenfassende Berechnungen (z.B. Mittelwerte, Location Entropy, KDE, Rasterisierung, Hotspot-Erkennung), Daten-Lokalität (z.B. Space Filling Curves und Geohash, Space-Time-Cubes, Clustering), Verarbeitung von Navigations- und Bewegungsdaten und weitere Themen an Beispielen diskutiert.

Praxisprojekt Topographie

Dozenten: Frank Thiemann, Malte Jan Schulze

>> Zur Veranstaltung im Stud.IP <<

Spatial and Spatiotemporal Statistics and Big Data

Dozenten: Prof. Dr. Phlipp Otto

Students are encouraged to critically analyse the performance of classical, statistical approaches for modelling spatial data under the presence of big and/or high-dimensional data. In this regard, students learn key concept of spatial and spatiotemporal statistics. Furthermore, an own simulation study is performed and described in a seminar paper.  

In a first part, important concepts of spatial and spatiotemporal statistics are introduced/repeated. In particular, the focus is on kriging and modelling spatial and spatiotemporal dependence by linear approaches, like autoregressive models. Further, we examine these approaches under the presence are large/big spatial data (incl. data streams) and discuss different approaches for reducing complexity and dimensionality. In the second half of the semester, all students work on a seminar paper assessing the performance of one of the introduced concepts for data with increasing size/complexity/dimensionality. Generally, these papers should include a small simulation study. There are individual obligatory meetings during the second half. The results of the seminar papers are presented in a colloquium.