Mitgliederversammlung DGSP mit Vortrag von PD. Dr. Sabine Wölk

Two important trends have emerged in information science in the last decades: artificial intelligence and quantum information. Amazing results have been achieved with artificial intelligence. For example, the algorithm AlphaGo learned to beat human players in the game Go by using reinforcement learning and playing against itself. As the field of artificial intelligence advances, the demand for fast and efficient algorithms increases.

Quantum information, on the other side, promises up to exponential speedups for certain computational tasks and enables speedups also in artificial intelligence. In addition, first quantum computer which can outperform classical ones despite their current restrictions are in reach.

In this talk, I will first give a short introduction to reinforcement learning, a branch of artificial intelligence. Then, I will introduce a hybrid learning agent which combines classical reinforcement learning with quantum search. Such a hybrid agent learns faster as a comparable classical agent as can be proven theoretically. However, to harvest this advantage also when solving real world problem, further  consideration have to be taken into account.

References

1. O. Sefrin and S. Wölk, Quant.Mach.Intell. 7, 52 (2025)
2. A.  Hamann  and S.  Wölk,  New J.  Phys. 24, 033044 (2022)
3. V. Saggio  et al., Nature  591, 229 (2021)