There are many possible topics in this area. A few examples are:
Manifold hypothesis. Look into network size and learning rate for learning a lower dimensional smoot or non-smooth manifold.
Convergence rates for DNNs
Regularization in training of DNNs, e.g. drop out.
Effect of initialisation and signal propagation in training DNNs
Function spaces for DNNs, in particular Barron spaces
Supervisors
Primary Supervisor: Fabian Dunker
Key qualifications and skills
Strong background in mathematical statistics and statistical learning theory.
Does the project come with funding
No - Student must be self-funded
Final date for receiving applications
Ongoing
How to apply
Apply by email to primary supervisor
Keywords
statistical learning
