TY  - Type of reference 
TI  - Experimental impact of (non-differentiable) image filters on image recognition residual networks 
AU  - Marc Fiammante 
AB  - This article describes the accuracy improvements observer by the injection of mathematically non-differentiable image filters As an introduction it provides a small state of the art of the techniques used in image recognition neural networks. 
DO  - 10.21494/ISTE.OP.2019.0359 
JF  - Cognitive Engineering 
KW  - Image recognition,  Neural Networks,  CNN,  differentiability,  wavelets, Reconnaissance d’image,  réseaux de neurones,  CNN,  filtres d’images,  dérivabilité,  ondelettes,  Keras,  
L1  - https://openscience.fr/IMG/pdf/iste_ingecog19v3n1_1.pdf 
LA  - en 
PB  - ISTE OpenScience 
DA  - 2019/04/16 
SN  - 2517-6978 
TT  - Amélioration de la reconnaissance d’image en utilisant des filtres d’image non réversibles avec des réseaux résiduels de neurones 
UR  - https://openscience.fr/Experimental-impact-of-non-differentiable-image-filters-on-image-recognition 
IS  - Issue 1 
VL  - 3 
ER  -