Frame Theory meets Neural Networks

This is my PhD project, funded by the DOC grant A 26355 and conducted at the Acoustics Research Institute (ARI), Vienna. The supervisors are Peter Balazs (ARI) and Martin Ehler (NuHAG).

The overall goal of this project is to make use of a frame-theoretic perspective to study the invertibility and stability of neural networks and propose architecture designs and regularizers that guarantee stable invertibility while being as efficient as possible.