Mitigating measurement errors in multiqubit experiments

Reducing measurement errors in multiqubit quantum devices is critical for performing any algorithm. Here we show how to mitigate by a classical postprocessing of the measured outcomes. Our techniques apply experiment where outcomes are used computing expected values observables. Two error-mitigation schemes presented based on tensor product and correlated Markovian noise models. Error rates parametrizing these models can be extracted from calibration data using simple formula. mitigation achieved applying inverse matrix probability vector that represents noisy measurement. The overhead, including number measurements cost postprocessing, exponential $\ensuremath{\epsilon}n$, $\ensuremath{\epsilon}$ maximum error rate $n$ qubits. We report experimental demonstration our methods IBM Quantum stabilizer graph states with $n\ensuremath{\le}12$ qubits entangled 20-qubit generated low-depth random Clifford circuits.