|
Cross-entropy reduction of data matrix with restriction on information capacity of projectors and their norms
Y. S. Popkovab, A. Y. Popkovb, Y. A. Dubnovbc a Institute of Control Sciences of RAS, Moscow
b Federal Research Center “Computer Science and Control” of RAS, Moscow
c National Research University Higher School of Economics, Moscow
Abstract:
We develop a new method of dimensionality reduction based on direct and inverse
projection of data matrix and calculation of projectors minimizing cross-entropy
functional. Concept of information capacity of matrix which is used as a restriction in a
problem of optimal reduction is introduced. We conduct a comparison of proposed
method with known ones based on binary classification.
Keywords:
dimensionality reduction, entropy, cross-entropy, classification.
Received: 15.01.2020 Revised: 12.03.2020 Accepted: 20.04.2020
Citation:
Y. S. Popkov, A. Y. Popkov, Y. A. Dubnov, “Cross-entropy reduction of data matrix with restriction on information capacity of projectors and their norms”, Matem. Mod., 32:9 (2020), 35–52; Math. Models Comput. Simul., 13:3 (2021), 382–394
Linking options:
https://www.mathnet.ru/eng/mm4212 https://www.mathnet.ru/eng/mm/v32/i9/p35
|
|