Nonnegative matrix factorization using a cross fuzzy entropy similarity measure
Assuming the structure Y=WH there are 2 versions of the code which have different updates of the H matrix:
- The version in the paper ICASSP_2016 which updates column-wise:
- FuzzyEn_Mean_HCol calcluates the nonnegative matrix factorization of an input matrix Y using a fuzzy entropy distance measure
- FuzzyEn_H_Update_Col_Mean calculates the current gradient of H column-wise
- One which updates row-wise:
- FuzzyEn_Mean_NMF calcluates the nonnegative matrix factorization of an input matrix Y using a fuzzy entropy distance measure
- FuzzyEn_H_Update_Mean calculates the current gradient of H row-wise
- Both use the following functions:
- FuzzyEn_W_Update_Mean calculates the current gradient of W
- CFuzzyEn_row calculates the cross fuzzy entropy between the rows of Y and WH
Requires the use of codegen to produce the mex files
ICASSP_2016 Fuzzy Entropy Based Nonnegative Matrix Factorization for Muscle Synergy Extraction