Skip to main content
. Author manuscript; available in PMC: 2017 Feb 1.
Published in final edited form as: IEEE Trans Pattern Anal Mach Intell. 2015 Sep 23;38(8):1640–1650. doi: 10.1109/TPAMI.2015.2481404

Fig. 5.

Fig. 5

The classification accuracies shown above are obtained using the same classification algorithm used in previous sections, but the data (features) now include additive Uniform noise. The x-axis specifies the length of support of the Uniform Distribution used in the noise model. The y-axis shows the classification accuracy. We see that realistic levels of noise do not have much of an effect on the derived algorithm. Note that in the 6DMG dataset the noise is added to both modalities and, hence, this corresponds to twice the amount of noise tested in other datasets.