The focus of my PhD is on developing deep learning methods for analysing historical document image datasets. As part of this, I work on classification, representation learning and training strategies that improve performance and reduce training time for deep neural networks. I'm also interested in ensuring reproducibility for deep learning experiments, so I helped develop and continue to work on DeepDIVA, a PyTorch based framework for reproducible deep learning experimentation.

In my PhD I focus on analyzing deep neural networks learning behaviour while working on computer vision tasks, such as historical documents image analysis. I am also a developper of DeepDIVA.

I am a PhD candidate working on the Graph-based Signature Verification project. In this project, graph-based methods are applied to the topic of signature verification. Additionally, we combine the graph-based approaches with deep learning methods in multiple classifier systems.