TY  - RPRT
TI  - Two SVDs produce more focal deep learning representations
AU  - Hinrich Schuetze
AU  - Christian Scheible
PY  - 2013
UR  - https://arxiv.org/abs/1301.3627
ID  - 1301.3627
ER  -