Thomas Wiatowski

Dr. sc. ETH Zurich, M.Sc. in Mathematics

Additional Information

You can find me on LinkedIn and ResearchGate.
Invited speaker at the Mathematics of Deep Learning workshop in Berlin, Germany, September 2017.
Co-organizer of the minisymposium Deep Neural Networks: Theory and Application at the Applied Inverse Problems (AIP) conference in Hangzhou, China, May 2017.

Note: Thomas Wiatowski is no longer with our group.

Publications

• Journal Papers and Manuscripts
  • Energy propagation in deep convolutional neural networks
    T. Wiatowski, P. Grohs, and H. Bölcskei, IEEE Transactions on Information Theory, Vol. 64, No. 7, pp. 4819-4842, July 2018.
  • A mathematical theory of deep convolutional neural networks for feature extraction
    T. Wiatowski and H. Bölcskei, IEEE Transactions on Information Theory, Vol. 64, No. 3, pp. 1845-1866, Mar. 2018.

• Conference, Symposium, and Workshop Papers
  • Deep structured features for semantic segmentation
    M. Tschannen, L. Cavigelli, F. Mentzer, T. Wiatowski, and L. Benini, Proc. of European Signal Processing Conference (EUSIPCO), pp. 61-65, Sept. 2017.
  • Topology reduction in deep convolutional feature extraction networks
    T. Wiatowski, P. Grohs, and H. Bölcskei, Proc. of SPIE (Wavelets and Sparsity XVII), San Diego, CA, USA, Vol. 10394, pp. 1039418:1-1039418:12, Aug. 2017, (invited paper).
  • Energy decay and conservation in deep convolutional neural networks
    P. Grohs, T. Wiatowski, and H. Bölcskei, Proc. of IEEE International Symposium on Information Theory (ISIT), Aachen, Germany, pp. 1356-1360, June 2017.
  • Heart sound classification using deep structured features
    M. Tschannen, T. Kramer, G. Marti, M. Heinzmann, and T. Wiatowski, Computing in Cardiology (CinC), Vancouver, Canada, pp. 565-568, Sept. 2016.
  • Deep convolutional neural networks on cartoon functions
    P. Grohs, T. Wiatowski, and H. Bölcskei, Proc. of IEEE International Symposium on Information Theory (ISIT), Barcelona, Spain, pp. 1163-1167, July 2016.
  • Discrete deep feature extraction: A theory and new architectures
    T. Wiatowski, M. Tschannen, A. Stanić, P. Grohs, and H. Bölcskei, Proc. of International Conference on Machine Learning (ICML), New York, USA, pp. 2149-2158, June 2016.
  • Deep convolutional neural networks based on semi-discrete frames
    T. Wiatowski and H. Bölcskei, Proc. of IEEE International Symposium on Information Theory (ISIT), Hong Kong, China, pp. 1212-1216, June 2015.

• Theses
  • Harmonic analysis of deep convolutional neural networks
    T. Wiatowski, Doctoral Thesis, ETH Zurich, Switzerland, Aug. 2017.
  • Kernel based image reconstruction from spherical Radon data
    T. Wiatowski, Master's Thesis, TU München, Germany, Nov. 2012.
  • Evolution of angular momentum expectation in quantum mechanics
    T. Wiatowski, Bachelor's Thesis, TU München, Germany, Aug. 2010.

Supervised Theses

• Diploma Theses
  • Noll, Andreas, "Topology Considerations in Scattering Networks," Fall semester 2017
    Supervisor(s): Thomas Wiatowski
  • Thandiackal, Kevin, "Wavelet-Based Convolutional Neural Networks for Reinforcement Learning," Spring semester 2016
    Supervisor(s): Michael Tschannen, Thomas Wiatowski
  • Stanić, Aleksandar, "Discrete Generalized Scattering Transform," Spring semester 2015
    Supervisor(s): Thomas Wiatowski
  • Cavigelli, Lukas, "Invariant Scattering Convolution Networks," Fall semester 2013
    Supervisor(s): Thomas Wiatowski, Céline Aubel

• Student Projects
  • Giacomuzzi, Sandro, "Rate of Energy Decay in Deep Convolutional Neural Networks," Fall semester 2017
    Supervisor(s): Thomas Wiatowski
  • Belkhayat, Kamil, "Statistical Arbitrage: Systematic Equity Pairs Trading," Spring semester 2017
    Supervisor(s): Thomas Wiatowski
  • Mentzer, Fabian, "Scene Labeling Using Deep Structured Features," Spring semester 2016
    Supervisor(s): Michael Tschannen, Lukas Cavigelli, Thomas Wiatowski, Michael Lerjen
  • Kühne, Jonas, "Scattering Networks for Scene Labeling," Fall semester 2015
    Supervisor(s): Michael Tschannen, Lukas Cavigelli, Thomas Wiatowski, Michael Lerjen
  • Geiger, Christian, "Feature Importance of Scattering Coefficients in Facial Landmark Detection," Fall semester 2015
    Supervisor(s): Michael Tschannen, Thomas Wiatowski
  • Thandiackal, Kevin, "Scattering Networks with Haar Wavelets," Spring semester 2015
    Supervisor(s): Thomas Wiatowski
  • Marti, Fabio, "Meshfree Approximation of High-Dimensional Data," Fall semester 2013
    Supervisor(s): Thomas Wiatowski

• Group Projects
  • Heinzmann, Matthias; Kramer, Thomas; Marti, Gian, "A Python Implementation for Deep Structured Feature Extraction," Spring semester 2016
    Supervisor(s): Thomas Wiatowski, Michael Tschannen

Talks

• Harmonic analysis of deep convolutional neural networks
  T. Wiatowski, Google Brain, Zurich, Switzerland, Nov. 2017, (invited talk).
• Deep convolutional neural networks from a harmonic analysis perspective
  T. Wiatowski, Mathematics of Deep Learning, Workshop organized by the Weierstrass Institute for Applied Analysis and Stochastics, Berlin, Germany, Sept. 2017, (invited talk).
• Topology reduction in deep convolutional feature extraction networks
  T. Wiatowski, Wavelets and Sparsity XVII, San Diego, USA, Aug. 2017, (invited talk).
• Harmonic analysis of deep convolutional neural networks
  T. Wiatowski, Applied Inverse Problems Conference, Hangzhou, China, May 2017, (invited talk).
• A mathematical theory of deep convolutional neural networks for feature extraction
  T. Wiatowski, Deep Learning: Theory and Practice, Workshop organized by the Max Planck Institute for Intelligent Systems, Donaueschingen, Germany, July 2016, (invited talk).
• Generic properties of scattering networks
  T. Wiatowski, Institute of Theoretical Information Technology, TU München, Germany, June 2016, (invited talk).
• Fast algebraic reconstruction for photoacoustic tomography
  T. Wiatowski, Institute of Computational Biology, Helmholtz Zentrum, Munich, Germany, Dec. 2013, (invited talk).
• Fast kernel-based reconstruction from spherical mean data
  T. Wiatowski, Applied Inverse Problems Conference, Daejeon, South Korea, July 2013, (invited talk).