Promotionsvortrag Physik: „Classical machine learning for problems in quantum computing“

Ankündigung des Promotionsvortrags von: Herrn Thomas Fösel

In recent years, machine learning and quantum computing have not only been individually among the most prospering fields in science and technology, but in particular also their intersection has established as a fruitful research topic. My PhD research is located here, with strong focus on applying classical machine learning to solve problems in quantum computing.
In our article Fösel et al. (2018) [PRX 8(3), 031084], we have shown how a neural-network-based “agent” can discover complete quantum-error-correction strategies, protecting a collection of qubits against noise; notably, these strategies require feedback adapted to observed measure-ment outcomes. In our article Fösel et al. (2021) [arXiv:2103.07585], we have proposed reinforce-ment learning for quantum circuit optimization, another central problem for quantum computing, and demonstrated how an “agent”, realized by a deep convolutional neural network, can autono-mously learn generic strategies to optimize arbitrary quantum circuits for used-defined quantum architectures and optimization targets.

In both cases, our reinforcement-learning-based approaches could find their success strategies from scratch, without any human guidance, but solely based on a given reward function. In this defense, I will discuss considered learning problems, key results and our solutions to technical challenges, and give a brief overview about my further contributions to this field.

(Vortrag auf Englisch)

Dem Vortrag schließt sich eine Diskussion von 15 Minuten an. Vortrag und Diskussion sind öffentlich. Diesen Verfahrensteilen folgt ein nicht öffentliches Rigorosum von 45 Minuten.