On the importance of normalisation layers in deep learning with piecewise linear activation units.

Zhibin Liao, Gustavo Carneiro

Browse the full WACV paper archive.