The paper discusses MIC in particular at least. They suggest that MIC sometimes overstates the strength of a noisy relationship, giving the example of a bivariate normal mixture (Figure 9). In that example MIC is 1, but the new measure is 0.48 or something, which seems more reasonable.
The paper discusses MIC in particular at least. They suggest that MIC sometimes overstates the strength of a noisy relationship, giving the example of a bivariate normal mixture (Figure 9). In that example MIC is 1, but the new measure is 0.48 or something, which seems more reasonable.