Fig. 4: MNIST data set.
From: Nearest centroid classification on a trapped ion quantum computer
(a) Ratio of distance calculated from experiment vs simulator, and (b) classification of different down-scaled MNIST data sets. From left to right, the data sets were (i) 0 and 1 (40 samples), (ii) 2 and 7 (40 samples), (iii) 0–3 (80 samples) and (iv) 0–9 (200 samples). The diamonds show the accuracy of classical Nearest Centroid. The error bars in (a) correspond to the standard error in the ratio of distances for each dataset.
