Modeling PCA approximation error via maximum-entropy random variables yields improved distance estimates over direct use of projected vectors.
A quantitative analy sis and performance study for similarity-search methods in high-d imensional spaces
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
stat.ML 1years
2019 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Improving the Accuracy of Principal Component Analysis by the Maximum Entropy Method
Modeling PCA approximation error via maximum-entropy random variables yields improved distance estimates over direct use of projected vectors.