Learning Solution Curves in Feedback-Based Quantum Algorithms

There is increasing interest in utilizing parameterized quantum circuits for solving problems of interest. Recently, feedback-based quantum algorithms have been introduced as optimization-free frameworks for this, where quantum circuit parameters are established layer-by-layer using a deterministic, measurement-based feedback law. The resulting set of parameters corresponds to what we refer to as solution curve. In this talk, I explore the prospect of sidestepping the typical measurement-based protocol and instead learning these solution curves with neural networks. I will present results exploring the quality of the predictions of neural networks and discuss potential future directions.