Featvect_matrix[:,f*self.NumCh+ch]=xor_bipolar(self.proj_mat_featuresCh[:,f*self.NumCh+ch],self.proj_mat_FeatVals[:,data[ch,f]])#!!! diffferent from binary
''' function to generate matrix of HD vectors using scale method with bits of randomization
scaleRand - every next vector is created by randomly flipping D/(numVec*scaleFact) elements - this way the further values vectors represent are, the less similar are vectors
print('Drop train:',(simpleAccuracy_train[-1,:]-simpleAccuracy_train[0,:])*100,' Drop test: ',(simpleAccuracy_test[-1,:]-simpleAccuracy_test[0,:])*100)