Comparison of the decoding accuracy of 'abstract' category information using different classifiers
The results below show a comparison of decoding accuracies for five
different classifier: correlation coefficient classifier
(CorrCoef, blue), regularized least squares (RLS, green),
nearest neighbor (NN, red), Gaussian Naive Bayes (NBG, cyan),
and Poisson Naive Bayes (NBP, purple) for ITC (upper figure) and
PFC (lower figure). These results are similar to those shown in
supplemental figure S2 except that here we are comparing more
classifiers and we are decoding 'abstract' category information (as was
done in Fig 3). As can be seen, the best performance is achieved
with the CorrCoef, NBG, and NBP classifiers, RLS achieves slightly
lower results and NN is by far the worst. However, for both areas
and all the classifiers (apart from NN which had very poor performance
overall), the general patterns of results is the same, which gives us
confidence that the classifier choice is not affecting the
conclusions drawn in this study.
Home