Predictive feedback for active noise canceling in superconducting quantum processors*

Superconducting qubits are amongst the most promising platforms for building near term practical quantum information processors. However, the coherence time (T2) of these qubits is negatively impacted in the presence of dephasing noise. In this work, we show how to efficiently estimate the noise spectrum experienced by the qubit, employ techniques from machine learning to predict the correlated noise in the future and use fast-feedback with an FPGA to cancel the noise. We propose using this estimate-predict-feedback sequence as a tool to increase gate fidelities while running a quantum circuit.