General Machine Learning

  • T. Krüger, D. Panknin, M. Braun, Fast Cross-Validation via Sequential Analysis, NIPS BigLearning Workshop, 2011 (abstract)

Machine Learning for Sensory Data and Control

  • D. Panknin, T. Krueger, M. Braun, K.-R. Müller, Siegmund Duell, Detecting changes in Wind Turbine Sensory Data, NIPS Workshop: Machine Learning for Sustainability, 2013 (abstract)

Social Media Analysis

Analysis of Twitter data; Impact Analysis; Big Data, Real-Time; Stream Mining.

  • M. Braun, M. Jugel, K.-R. Müller, Real-time social media analysis with TWIMPACT, Demonstration at NIPS 2011. (abstract)
  • F. Bießmann, F. Meinecke, M. Jugel, M. Braun, Online CCA for Realtime Impact Analysis of Social Media Data, NIPS Workshop on "Algorithmic and Statistical Approaches for Large Social Networks", 2012 (abstract)
  • F. Bießmann, J.-M. Papaioannou, A. Harth, M. Jugel, K.-R. Müller, M. Braun , Quantifying Spatiotemporal Dynamics of Twitter Replies to News Feeds, Proceedings of the IEEE Workshop on Machine Learning for Signal Processing, 2012 (abstract)
  • F. Bießmann, J.-M. Papaioannou, M. Braun, A. Harth, Canonical Trends: Detecting Trend Setters in Web Data, Proceedings of the International Conference of Machine Learning, 2012 (abstract)

(Deep) Neural Networks

Studying how representations are built in (deep) neural networks using the techniques developed for kernel methods.

  • G. Montavon, M. Braun, K.-R. Müller, Deep Boltzmann Machines as Feed-Forward Hierarchies, International Conference on Artificial Intelligence and Statistics (AISTATS), 2012 (abstract)
  • G. Montavon, M. Braun, K.-R. Müller, Importance of Cross-Layer Cooperation for Learning Deep Feature Hierarchies, NIPS Workshop on Deep Learning and Unsupervised Feature Learning, 2011 (abstract)
  • G. Montavon, M. Braun, K.-R. Müller, Kernel Analysis of Deep Networks, Journal of Machine Learning Research (JMLR), 12, pp. 2563-2581, 2011 (abstract)
  • G. Montavon, M. Braun, K.-R. Müller, Layer-wise analysis of deep networks with Gaussian kernels, Advances in Neural Information Processing Systems 23 (NIPS 2010), pp. 1678-1686, 2010 (abstract)

Machine Learning Open Source Software

Kernel Methods

Eigenstructure of the Kernel Matrix; Kernel PLS; Kernel Methods

  • G. Montavon, M. Braun, T. Krueger, K.-R. Müller, Analyzing Local Structure in Kernel-Based Learning: Explanation, Complexity, and Reliability Assessment , IEEE Signal Processing Magazine, Volume 30:4, p. 62-74, July 2013 (abstract)
  • N. Krämer, M. Sugiyama, M.L. Braun, Lanczos Approximations for the Speedup of Kernel Partial Least Squares Regression , JMLR Workshop and Conference Proceedings, Volume 5: AISTATS 2009 (abstract)
  • M.L. Braun, J. Buhmann, K.-R. Müller, On Relevant Dimensions in Kernel Feature Spaces, JMLR 9(Aug):1875--1908, 2008. (software) (more information) (abstract)
  • N. Krämer, M.L. Braun, Kernelizing PLS, Degress of Freedom, and Efficient Model Selection, ICML 2007 (abstract)
  • M. L. Braun, Accurate bounds for the eigenvalues of the kernel matrix, Journal of Machine Learning Research, Vol. 7(Nov) 2303-2328, 2006 (abstract)
  • M.L. Braun, J. Buhmann, K.-R. Müller, Denoising and Dimension Reduction in Feature Space, Advances in Neural Information Processing Systems 19 (NIPS 2006), B. Schölkopf, J. Platt, and T. Hoffman (eds.), MIT Press, Cambridge, MA, 185-192, 2007 (abstract)
  • M.L. Braun, T. Lange, J. Buhmann, Model Selection in Kernel Methods based on a Spectral Analysis of Label Information , DAGM 2006, LNCS 4174, pp. 344-353 (abstract)
  • M. L. Braun, Spectral Properties of the Kernel Matrix and their Relation to Kernel Methods in Machine Learning , Ph.D. thesis, University of Bonn, 2005. Published electronically. (abstract)

Brain Computer Interface

Detecting conditions of high mental and auditory workload based on EEG recordings under real-world conditions.

Stability Based Model Selection

A general framework for estimating the right number of clusters based on the stability of the clustering solutions.

Noisy TSP

Applying methods from statistical physics to robustly sample good solutions for combinatorial optimization problems like the travelling salesman problem.

  • M. Braun, J. Buhmann, The Noisy Euclidian Traveling Salesman Problem and Learning, in: T. G. Diettrich, S. Becker, Z. Ghahramani (Eds.), Advances in Information Processing Systems 14, pp. 251-258, MIT Press (abstract)

Retina Implant

From Mar. 1997 to Sept. 1999, I worked as a lab assitant in the Retina Implant project (which was a registered project of the EXPO2000) with Michael Becker. The goal was to construct a retinal prosthesis for people with macula degeneration.