Quantum Measurement Discrimination using Cumulative Distribution Functions

Quantum measurement is one of the critical steps in quantum computing that determines the probabilities associated with qubit states after conducting several circuit ex-ecutions and measurements. As a mesoscopic quantum system, real quantum computers are prone to noise. Therefore, a major challenge in quantum measurement is how to correctly inter-pret the noisy results of a quantum computer. While there are promising classification based solutions, they either produce incorrect results (misclassify) or require many measurements (expensive). In this paper, we present an efficient technique to estimate a qubit's state through analysis of probability distributions of post-measurement data. Specifically, we estimate the state of a qubit using cumulative distribution functions to compare the measured distribution of a sample with the distributions of basis states
$\vert 0\rangle$
and
$\vert 1\rangle$
. Our experimental results demonstrate a drastic reduction (78%) in single qubit readout error. It also provides significant reduction (12%) when used to boost existing multi-qubit discriminator models.