Conceptual Foundations for Machine Learning in Engineering Using the Example of Tailored Laser Pulse Forming

DASHH Doctoral Researcher: Utsa Chattopadhyay

Supervisors: Dr. Christoph Heyl (DESY), Dr. Henrik Tünnermann (DESY), Prof. Nihat Ay (TUHH)

Machine Learning and neural networks offer tremendous opportunities in engineering and science, reaching from fault detection to modelling physical systems for technological innovations. This has been demosntrated by the many applications within the TUHH initiative "Machine Learning in Engineering" (MLE). However, despite these demonstrations, the conceptual and theoretical foundations behind machine learning-based approaches to engineering problems are largely missing.
This project will contribute to these foundations within a far-reaching application setting, thereby perfectly matching the core vision of DASHH. In particular, we will investigate the feasibility of employing neural networks to speed up optimization problems using the example of a challenging application-oriented problem from nonlinear photonics while striving to derive foundational principles underlying machine learning approaches. By synergetically combining the expertise from data sciences and laser physics, the project outcome can thus be two-fold: It can lay the ground work to foundations for engineering-oriented machine learning approaches while solving a key challenge for laser physics exploring an only recently proposed method promising the production of intense tailored light pulses within novel wavelength regimes.