Fig. 1From: Unsupervised machine learning applied to scanning precession electron diffraction dataSPED data from a GaAs nanowire and virtual dark-field images formed by plotting the intensity within the disks marked around \(\lbrace 111\rbrace\) reflections, as a function of probe position. a Without precession and b with 35 mrad precession. Diffraction pattern and VDF image scale bars are common to all subfigures and measure 1 Ã…\(^{-1}\) and 150Â nm respectively. The approximate position of the carbon film is indicated by the red dashed lineBack to article page