Thesis, where we consider a stochastic control problem and try to solve it using the signature method.

Thesis, where we extend and test KEdge, an interpretable-by-design approach for graph neural networks, and compare it to gradient-based attribution techniques.