%0 Conference Proceedings
%T Attenuating Catastrophic Forgetting by Joint Contrastive and Incremental Learning
%+ Equipe Models and AlgoriThms for pRocessIng and eXtracting information (Lab-STICC_MATRIX)
%+ École Nationale Supérieure de Techniques Avancées Bretagne (ENSTA Bretagne)
%+ Naval Group Research [Bouguenais]
%+ FrenCh austRalian labOratory for humanS/autonomouS agents teamING  (CROSSING)
%+ Equipe Robot interaction, Ambient system, Machine learning, Behaviour, Optimization (Lab-STICC_RAMBO)
%+ Département Informatique (IMT Atlantique - INFO)
%A Ferdinand, Quentin
%A Clement, Benoit
%A Oliveau, Quentin
%A Le Chenadec, Gilles
%A Papadakis, Panagiotis
%< avec comité de lecture
%( 2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW)
%B IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)
%C New-Orleans, United States
%8 2022-06-19
%D 2022
%R 10.1109/CVPRW56347.2022.00423
%Z Engineering Sciences [physics]/AutomaticConference papers
%X In class incremental learning, discriminative models are trained to classify images while adapting to new instances and classes incrementally. Training a model to adapt to new classes without total access to previous class data, however, leads to the known problem of catastrophic forgetting of the previously learnt classes. To alleviate this problem, we show how we can build upon recent progress on contrastive learning methods. In particular, we develop an incremental learning approach for deep neural networks operating both at classification and representation level which alleviates forgetting and learns more general features for data classification. Experiments performed on several datasets demonstrate the superiority of the proposed method with respect to well known state-of-the-art methods.
%G English
%2 https://hal-ensta-bretagne.archives-ouvertes.fr/hal-03784379/document
%2 https://hal-ensta-bretagne.archives-ouvertes.fr/hal-03784379/file/Ferdinand_Attenuating_Catastrophic_Forgetting_by_Joint_Contrastive_and_Incremental_Learning_CVPRW_2022_paper.pdf
%L hal-03784379
%U https://hal-ensta-bretagne.archives-ouvertes.fr/hal-03784379
%~ UNIV-BREST
%~ INSTITUT-TELECOM
%~ ENSTA-BRETAGNE
%~ CNRS
%~ UNIV-UBS
%~ ENSTA-BRETAGNE-STIC
%~ ENIB
%~ LAB-STICC
%~ TDS-MACS
%~ IMTA_INFO
%~ LAB-STICC_IMTA
%~ IMT-ATLANTIQUE
%~ PRACOM
%~ INSTITUTS-TELECOM
%~ LAB-STICC_MATRIX_IMTA
%~ LAB-STICC_RAMBO_IMTA
%~ LAB-STICC_MATRIX
%~ LAB-STICC_RAMBO
%~ LAB-STICC_DMID
%~ LAB-STICC_INTERACTION