Zur Hauptnavigation / To main navigation

Zur Sekundärnavigation / To secondary navigation

Zum Inhalt dieser Seite / To the content of this page

  • EN

Hauptnavigation / Main Navigation

Sekundärnavigation / Secondary navigation

Seminar Learning

Inhaltsbereich / Content

Seminar Learning

 

+=

Data, data, data...

One huge, but exciting challenge of the 21th century posed to

mathematicians is the deluge of data, caused by

  • Telecommunication
  • Medicine
  • Genomics
  • Social Networks
  • Astronomy
  • ...

Various types of data

  • 1D functions: Signals
  • 2D functions: Images
  • 3D functions: Videos
  • >= 1000D function: High-Dimensional Data
  • Functions on manifolds and manifold-valued functions
  • Functions on graphs

Key approaches

  • Applied Harmonic Analysis – Finding suitable structured representations!
  • Deep Neural Networks – Learning representations, in particular, for classification
  • Compressed Sensing – Acquiring only the compressed part of data
    This can be used to solve linear inverse problems in general.
  • Machine Learning – Find patterns in data
  • Spectral Theory (on Graphs) – Analyze the spectrum of data

Schedule

The seminar takes place in the lecture room IBZ-302. The schedule was fixed as follows

 

DateTalk (Speaker)
02.11.17, 3.30pmSVM (C. Hertrich)
09.11.17, 3.30pm Kernel-Trick (R. Rosandi) and RKHS (M. Winkler)
16.11.17, 3.30pmNeural Networks and Backpropagation (B. Schmitt)
30.11.17, 3.30pmStatistical Learning (P. Capelli)
07.12.17, 3.30pm(L)ISTA (M. Brusius, S. Neumayer)
11.01.18, 3.30pm Convolutional Neural Networks (J. Settelmaier)
12.01.18, 11.45pm Convolutional Neural Networks II (M. Wiese)

The image of the painting “Composition 8” by Wassliy Kandinsky is public domain and available in the wikimedia.