Team members wanted (m/f/d)

If you are close to finishing your master thesis (or have recently finished) and are looking for a PhD position in the area of mathematics of deep learning, please write an email to felix.voigtlaender(at)ku.de!

Among others, I am offering the following topics:

• Existence of adversarial examples for neural networks trained by stochastic gradient descent.
  Preferred background: Probabilistic methods and concentration inequalities, in particular for random matrices.
• Learning bounds in an adversarial setting.
  Preferred background: Knowledge of PAC learning and VC dimension. Preferably also an interest in algebraic geometry.
• Computability of neural network optimization problems.
  Preferred background: Good knowledge of real and functional analysis, preferably also knowledge regarding sampling or information-based complexity.