Logo des Repositoriums

Auflistung nach Autor:in "Huber, Marco F."

1 - 3 von 3
  • Konferenzbeitrag
    Big data architecture for the semantic analysis of complex events in manufacturing
    (Informatik 2016, 2016) Huber, Marco F.; Voigt, Martin; Ngomo, Axel-Cyrille Ngonga
  • Konferenzbeitrag
    On Sensor Scheduling in Case of Unreliable Communication
    (Informatik 2007 – Informatik trifft Logistik – Band 2, 2007) Huber, Marco F.; Stiegeler, Eric; Hanebeck, Uwe D.
    This paper deals with the linear discrete-time sensor scheduling problem in unreliable communication networks. The sensor scheduling problem, where one sensor from a sensor network is selected for performing a measurement at a specific time instant so that the estimation errors are minimized, can be solved off-line by extensive tree search, in case an error-free communication is assumed. The main contribution of the proposed scheduling approach is to introduce a prioritization list for the sensors that leads to a minimization of the estimation error by selecting the most beneficial sensor even in case of unreliable communication. To lower the computational demand for the priority list calculation, an optimal pruning approach is introduced.
  • Konferenzbeitrag
    Superficial Gaussian mixture reduction
    (INFORMATIK 2011 – Informatik schafft Communities, 2011) Huber, Marco F.; Krauthausen, Peter; Hanebeck, Uwe D.
    Many information fusion tasks involve the processing of Gaussian mixtures with simple underlying shape, but many components. This paper addresses the problem of reducing the number of components, allowing for faster density processing. The proposed approach is based on identifying components irrelevant for the overall density's shape by means of the curvature of the density's surface. The key idea is to minimize an upper bound of the curvature while maintaining a low global reduction error by optimizing the weights of the original Gaussian mixture only. The mixture is reduced by assigning zero weights to reducible components. The main advantages are an alleviation of the model selection problem, as the number of components is chosen by the algorithm automatically, the derivation of simple curvature-based penalty terms, and an easy, efficient implementation. A series of experiments shows the approach to provide a good trade-off between quality and sparsity.