Special issue on machine learning in finance

Artificial intelligence (AI) and machine learning (ML) certainly range among the topics that have not only moved to the center of public discourse but also the center of many research fields. One of these is Mathematical Finance, where AI and ML approaches become more and more critical, both for the application of data-driven methods to financial problems as well as for the understanding of their mathematical and probabilistic foundations. To create an international platform for talks on the most recent advances in this area and for an extensive exchange of ideas during the COVID-19 pandemic, we decided in March 2021 to set up a world online seminar series on “Machine Learning in Finance” (see https://sites.google.com/view/mlfinance/). We could gain many leading scientists, practitioners as well as highly promising young researchers as our speakers covering topics related to all kinds of (deep) learning techniques in finance. This included, in particular, the analysis of training algorithms for supervised learning, reinforcement learning and GANs, connections to game theory, signature methods, model calibration, market-making, financial networks, dynamic risk assessment, as well as novel deep learning models capturing geometric structure and causality. The current special issue grew out of this seminar series, and we invited the speakers of the first year to submit their works. This resulted in nine highly insightful and innovative papers that cover a broad spectrum of themes and whose content is described briefly below. We choose here the same order as the order of speakers in the seminar series.