Paper published in a book (Scientific congresses and symposiums)
An Effective Hit-or-Miss Layer Favoring Feature Interpretation as Learned Prototypes Deformations
Deliège, Adrien; Cioppa, Anthony; Van Droogenbroeck, Marc
2019In Thirty-Third AAAI Conference on Artificial Intelligence
Peer reviewed
 

Files


Full Text
Deliege2019AnEffective.pdf
Author preprint (873.49 kB)
Download
Annexes
README.MD
Publisher postprint (2.85 kB)
Download
Codes.zip
Publisher postprint (6.64 kB)
Download
Videos.zip
Publisher postprint (42.48 MB)
Download
Deliege2019AnEffective.bib
Publisher postprint (651 B)
BiBTeX entry
Download

All documents in ORBi are protected by a user license.

Send to



Details



Keywords :
deep learning; artificial intelligence; neural network; classification; capsule; Hitnet; Hit-or-miss layer; feature interpretation; centripetal loss; prototype
Abstract :
[en] Neural networks designed for the task of classification have become a commodity in recent years. Many works target the development of more effective networks, which results in a complexification of their architectures with more layers, multiple sub-networks, or even the combination of multiple classifiers, but this often comes at the expense of producing uninterpretable black boxes. In this paper, we redesign a simple capsule network to enable it to synthesize class-representative samples, called prototypes, by replacing the last layer with a novel Hit-or-Miss layer. This layer contains activated vectors, called capsules, that we train to hit or miss a fixed target capsule by tailoring a specific centripetal loss function. 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. We show that our network, named HitNet, is able to reach better performances than those reproduced with the initial CapsNet on several datasets, while allowing to visualize the nature of the features extracted as deformations of the prototypes, which provides a direct insight into the feature representation learned by the network.
Research center :
Montefiore Institute of Electrical Engineering and Computer Science - Montefiore Institute ; Telim
Disciplines :
Electrical & electronics engineering
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
Language :
English
Title :
An Effective Hit-or-Miss Layer Favoring Feature Interpretation as Learned Prototypes Deformations
Publication date :
February 2019
Event name :
Workshop on Network Interpretability for Deep Learning at the Thirty-Third AAAI Conference on Artificial Intelligence
Event organizer :
AAAI
Event place :
Honolulu, United States
Event date :
27 Jan 2019 to 1 Feb 2019
Audience :
International
Main work title :
Thirty-Third AAAI Conference on Artificial Intelligence
Peer reviewed :
Peer reviewed
Name of the research project :
DeepSport
Funders :
DGTRE - Région wallonne. Direction générale des Technologies, de la Recherche et de l'Énergie [BE]
Commentary :
Accepted for AAAI workshop on "Network interpretability for deep learning" : https://aaai.org/Conferences/AAAI-19/ws19workshops/#ws10.
Available on ORBi :
since 29 November 2018

Statistics


Number of views
389 (93 by ULiège)
Number of downloads
290 (35 by ULiège)

Bibliography


Similar publications



Contact ORBi