A Cumulants-Based Human Brain Decoding

Algorithm 1

Optimal prediction algorithm.

(i)	Data Transformation.
(a)	Estimated values of the voxels are found using the General linear model.
(b)	The significant brain area is found and shown based on the highest T-value.
(ii)	Data Selection
(a)	The significant features are selected based on p-values.
(b)	Set several selected features having lower p-values.
(iii)	Feature Extraction
(a)	Extract the HOCs from data using equation (5) , where N is number of selected features.
(iv)	Likelihood Ratio Test
(a)	Do the estimation of the densities for different classes using the Kernel density estimator.
(b)	The density-based score fusion is done using likelihood ratio-based score fusion
(c)	The estimation of the score of the class is done using the likelihood ratio test using equation (6).