Dr. rer. nat.

Hecker, Dirk

Filters

2
Reset filters

Settings
Now showing 1 - 3 of 3
  • Publication
    Informed Machine Learning for Industry
    ( 2019)
    Bauckhage, Christian
    ;
    Schulz, Daniel
    ;
    Hecker, Dirk
    Deep neural networks have pushed the boundaries of artificial intelligence but their training requires vast amounts of data and high performance hardware. While truly digitised companies easily cope with these prerequisites, traditional industries still often lack the kind of data or infrastructures the current generation of end-to-end machine learning depends on. The Fraunhofer Center for Machine Learning therefore develops novel solutions which are informed by expert knowledge. These typically require less training data and are more transparent in their decision-making processes.
  • Publication
    Detecting Mobility Patterns with Stationary Bluetooth Sensors: A real-world Case Study
    ( 2015)
    Müller, Marc
    ;
    Schulz, Daniel
    ;
    Mock, Michael
    ;
    Hecker, Dirk
    A Bluetooth sensor network was built up in the city of Bonn to measure Bluetooth MAC-addresses. The results of the acquired data are separated on a macro level and mobility patterns. We have collected nearly 5 million data points from 14 distinct stationary sensors over a period of 1 month and recognized over 85.000 unique devices. We show that the data is sufficiently dense to detect commuter patterns based on a Fourier analysis. In addition, we discuss limitations found in the dataset and present lessons learned.
  • Publication
    Pedestrian flow prediction in extensive road networks using biased observational data
    ( 2008)
    Scheider, Simon
    ;
    May, Michael
    ;
    Rösler, Roberto
    ;
    Schulz, Daniel
    ;
    Hecker, Dirk
    In this paper, we discuss an application of spatial data mining to predict pedestrian flow in extensive road networks using a large biased sample. Existing out-of-the-box techniques are not able to appropriately deal with its challenges and constraints, in particular with sample selection bias. For this purpose, we introduce s-knn-apriori, an efficient nearest neighbor based spatial mining algorithm that allows prior knowledge and deductive models to be included in a straightforward and easy way.