Article (Scientific journals)
HitNet: a neural network with capsules embedded in a Hit-or-Miss layer, extended with hybrid data augmentation and ghost capsules
Deliège, Adrien; Cioppa, Anthony; Van Droogenbroeck, Marc
2018In arXiv, 1806.06519
 

Files


Full Text
Deliege2018HitNet.pdf
Publisher postprint (1.37 MB)
HitNet: a neural network with capsules embedded in a Hit-or-Miss layer, extended with hybrid data augmentation and ghost capsules
Download
Annexes
HitNet.zip
Publisher postprint (6.64 kB)
Python source code
Download
logo-HitNet.png
Publisher postprint (656.21 kB)
HitNet logo
Download

All documents in ORBi are protected by a user license.

Send to



Details



Keywords :
Artificial intelligence; Neural network; Deep learning; Classification; Capsule; Centripetal loss; Ghost capsule; Hit-or-Miss layer; HitNet
Abstract :
[en] Abstract Neural networks designed for the task of classification have become a commodity in recent years. Many works target the development of better networks, which results in a complexification of their architectures with more layers, multiple sub-networks, or even the combination of multiple classifiers. In this paper, we show how to redesign a simple network to reach excellent performances, which are better than the results reproduced with CapsNet on several datasets, by replacing a layer with a Hit-or-Miss layer. This layer contains activated vectors, called capsules, that we train to hit or miss a central capsule by tailoring a specific centripetal loss function. We also show how our network, named HitNet, is capable of synthesizing a representative sample of the images of a given class by including a reconstruction network. This possibility allows to develop a data augmentation step combining information from the data space and the feature space, resulting in a hybrid data augmentation process. In addition, we introduce the possibility for HitNet, to adopt an alternative to the true target when needed by using the new concept of ghost capsules, which is used here to detect potentially mislabeled images in the training data.
Research center :
Montefiore Institute - Montefiore Institute of Electrical Engineering and Computer Science - ULiège
Telim
Disciplines :
Computer science
Author, co-author :
Deliège, Adrien  ;  Université de Liège - ULiège > Dép. d'électric., électron. et informat. (Inst.Montefiore) > Télécommunications
Cioppa, Anthony  ;  Université de Liège - ULiège > Dép. d'électric., électron. et informat. (Inst.Montefiore) > Télécommunications
Van Droogenbroeck, Marc   ;  Université de Liège - ULiège > Dép. d'électric., électron. et informat. (Inst.Montefiore) > Télécommunications
 These authors have contributed equally to this work.
Language :
English
Title :
HitNet: a neural network with capsules embedded in a Hit-or-Miss layer, extended with hybrid data augmentation and ghost capsules
Publication date :
18 June 2018
Journal title :
arXiv
Volume :
1806.06519
Name of the research project :
DeepSport
Funders :
DGTRE - Région wallonne. Direction générale des Technologies, de la Recherche et de l'Énergie [BE]
Available on ORBi :
since 19 June 2018

Statistics


Number of views
166 (41 by ULiège)
Number of downloads
237 (20 by ULiège)

Bibliography


Similar publications



Contact ORBi