Different architectures for neural ordinary differential equations

Machine learning has been getting more and more important during the last decades. One of the most important tools in machine learning are neural networks. A rather modern approach of constructing a neural network is using a neural ordinary differential equation (or neural ODE). Here, the idea is to construct a neural network which can be evaluated by (numerically) solving an ODE. Neural ODEs are a powerful tool to solve many different machine learning problems. However, it is not so easy to construct a fitting neural ODE model in practice. In the thesis, some basic ways of constructing a neural ODE are explored.

Autorentext
Elias Walder, BSc, Dipl.-Ing., studied at the University of Innsbruck, Faculty of Mathematics, Computer Science and Physics. His academic focus lies in applied mathematics.