Classification using Distinguishability Measure
In this page, we use distinguishablity measure to come up with the notion of "versatile" and "non-informative" voxels. Then we perform classification using both type of voxels alone just to see how each type performs. Plotting the accuracy curve obtained when using versatile voxels (V>=V_treshold) and non-informative (V<V_treshold) voxels. The figures below show the voxels from VT cortex (top, 577 voxels) and from the whole brain (bottom, 43193 voxels) respectively.