When superconducting transmon qubits are measured with a readout pulse, the raw readout signal is demodulated, and results appear as clouds on the IQ plane, with one point in the cloud representing one time the quantum state was prepared and measured.
Quantum experiments typically require computing expectation values of measurement operators, and that entails turning the results of each shot into a 0 or 1 and performing statistical analysis on the collections of 0s and 1s.
For example, if I was doing quantum state tomography, then I would end up using the 1s and 0s to do a maximum likelihood estimation. The thing is, some measurement results give more information than others: a blue dot that's far away from the red cloud gives a 0 with more confidence than one near the overlap region.
Question: Has there been any effort to try to process the data collectively rather that individually? That is, not reducing all point to 1s and 0s and doing statistics, but processing all the points together using their positions relative to all other points in the cloud.
