Thomas Allard Research Assistant

Address

ETH Zürich
Chair for Mathematical Information Science
Sternwartstr. 7
CH-8092 Zürich
Switzerland

Room: ETF E 118

Tel.: +41 44 632 45 76

Email: tallard (add )

Publications

• Journal Papers and Manuscripts
  • Entropy of compact operators with applications to Landau-Pollak-Slepian theory and Sobolev spaces
    T. Allard and H. Bölcskei, Applied and Computational Harmonic Analysis, June 2024, submitted.
  • Ellipsoid methods for metric entropy computation
    T. Allard and H. Bölcskei, Constructive Approximation, May 2024, submitted.

Supervised Theses

• Diploma Theses
  • Schneiderbauer, Maximilian, "Metric Entropy Optimality of Continuous-Time RNNs for Learning Dynamical Systems," Fall semester 2023
    Supervisor(s): Thomas Allard
  • Häberle, Konstantin, "Learning Rate Scheduling for Stochastic Gradient Descent," Spring semester 2021
    Supervisor(s): Thomas Allard

• Student Projects
  • Hultberg, Bror, "Metric Entropy of Hypoelliptic Operators," Spring semester 2024
    Supervisor(s): Thomas Allard, Clemens Hutter
  • Schneiderbauer, Maximilian, "On the Metric Entropy of Dynamical Systems," Spring semester 2023
    Supervisor(s): Thomas Allard
  • Waber, Jivan, "Metric Entropy of Pseudodifferential Operators," Fall semester 2022
    Supervisor(s): Thomas Allard
  • Druenne, Hugo, "Approximation of Dynamical Systems by Recurrent Neural Networks," Spring semester 2022
    Supervisor(s): Thomas Allard
  • Tombak, Güney , "Noise in Stochastic Gradient Descent with respect to Expected Loss," Fall semester 2021
    Supervisor(s): Thomas Allard
  • Ayari, Rayen, "On the News Categorization," Fall semester 2021
    Supervisor(s): Thomas Allard
  • Häberle, Konstantin, "Random Perturbations Theory for Stochastic Gradient Descent," Spring semester 2020
    Supervisor(s): Thomas Allard
  • Clarysse, Jacob, "On Parameters of the Expressivity of Neural Networks," Spring semester 2020
    Supervisor(s): Thomas Allard

• Bachelor Theses
  • Lahmann, Pablo, "Stability of Simple Neural Network Architectures," Fall semester 2021
    Supervisor(s): Thomas Allard
  • Iliadis, Afroditi, "SGD Learns Over-parametrized Networks that Provably Generalize on Linearly Separable Data," Spring semester 2021
    Supervisor(s): Thomas Allard

• Group Projects
  • Iliadis, Afroditi; Heidtmann, Isabel, "Optimization and Regularization Methods for Neural Networks: A Literature Review," Fall semester 2020
    Supervisor(s): Thomas Allard