Quantum reservoir computing (QRC) is a supervised machine learning algorithm that harnesses the input-driven dynamics of a complex quantum system (viz. `reservoir’) to solve temporal tasks, by leveraging the system’s intrinsic memory of a few past inputs. Compared to conventional neural networks, reservoir computing provides much simpler and faster training. In this talk, following a brief outline of QRC, I will present some intriguing aspects of QRC using a hybrid qubit-boson system governed by the Jaynes–Cummings Hamiltonian and its dispersive limit. This includes an analysis of their memory capacity across parameter regimes, and their performance in predicting chaotic time-series data.
