Geodesic
Space semantic aware loss function for embedding creation in case of transaction data
Journal of the Belarusian State University. Mathematics and Informatics, Tome 1 (2022), pp. 97-102

Voir la notice de l'article provenant de la source Math-Net.Ru

Transaction data are the most popular data type of bank domain, they are often represented as sparse vectors with a large number of features. Using sparse vectors in deep learning tasks is computationally inefficient and may lead to overfitting. Аutoencoders are widely applied to extract new useful features in a lower dimensional space. In this paper we propose to use a novel loss function based on the metric that estimates the quality of mapping the semantic structure of the original tabular data to the embedded space. The proposed loss function allows preserving the item relation structure of the original space during the dimension reduction transformation. The obtained results show the improvement of the resulting embedding properties while using the combination of the new loss function and the traditional mean squared error one.
Keywords: data; embedding; vector; loss function; autoencoder. 
@article{BGUMI_2022_1_a9,
     author = {M. E. Vatkin and D. A. Vorobey and M. V. Yakovlev and M. G. Krivova},
     title = {Space semantic aware loss function for embedding creation in case of transaction data},
     journal = {Journal of the Belarusian State University. Mathematics and Informatics},
     pages = {97--102},
     publisher = {mathdoc},
     volume = {1},
     year = {2022},
     language = {en},
     url = {http://geodesic.mathdoc.fr/item/BGUMI_2022_1_a9/}
}
TY  - JOUR
AU  - M. E. Vatkin
AU  - D. A. Vorobey
AU  - M. V. Yakovlev
AU  - M. G. Krivova
TI  - Space semantic aware loss function for embedding creation in case of transaction data
JO  - Journal of the Belarusian State University. Mathematics and Informatics
PY  - 2022
SP  - 97
EP  - 102
VL  - 1
PB  - mathdoc
UR  - http://geodesic.mathdoc.fr/item/BGUMI_2022_1_a9/
LA  - en
ID  - BGUMI_2022_1_a9
ER  - 
%0 Journal Article
%A M. E. Vatkin
%A D. A. Vorobey
%A M. V. Yakovlev
%A M. G. Krivova
%T Space semantic aware loss function for embedding creation in case of transaction data
%J Journal of the Belarusian State University. Mathematics and Informatics
%D 2022
%P 97-102
%V 1
%I mathdoc
%U http://geodesic.mathdoc.fr/item/BGUMI_2022_1_a9/
%G en
%F BGUMI_2022_1_a9
M. E. Vatkin; D. A. Vorobey; M. V. Yakovlev; M. G. Krivova. Space semantic aware loss function for embedding creation in case of transaction data. Journal of the Belarusian State University. Mathematics and Informatics, Tome 1 (2022), pp. 97-102. http://geodesic.mathdoc.fr/item/BGUMI_2022_1_a9/