Sorry, you need to enable JavaScript to visit this website.

facebooktwittermailshare

DISCRIMINATIVE FEATURES FOR INCREMENTAL LEARNING CLASSIFIER

Abstract: 

An important problem in artificial intelligence is to develop an ef-
ficient system that can adapt to new knowledge in an incremen-
tal manner without forgetting previously learned knowledge. Al-
though Convolutional Neural Networks (CNNs) are good at learn-
ing strong classifier and discriminative features, CNNs can not per-
form well in incremental classifier learning due to the catastrophic
forgetting problem in the retraining process. In this paper, we pro-
pose a novel yet extremely simple approach to enhance the discrim-
inative property of features for incremental classifier learning. We
build a network for the universal feature space in which a group of
image classes have intra-class compactness and inter-class separa-
bility. And, we model each incremental class to have a maximum
margin from the rest of the models in universal space. Experiments
are conducted on CIFAR-100 dataset and IMage Database for Con-
text Aware Advertisement (IMDB-CAA) we collected. The results
demonstrate the superiority of our approach, improving performance
on CIFAR-100 dataset over state-of-the-art incremental learning sys-
tems. Furthermore, experiments on few-short incremental learning
setting show very promising performance although we use only 4%
of training samples on CIFAR-100 dataset.

up
0 users have voted:

Comments

DISCRIMINATIVE FEATURES FOR INCREMENTAL LEARNING CLASSIFIER

DISCRIMINATIVE FEATURES FOR INCREMENTAL LEARNING CLASSIFIER

Paper Details

Authors:
Balaji Nataraj, Xie Shudong, Li Yiqun, Lin Dongyun, Dong Sheng
Submitted On:
20 September 2019 - 6:33am
Short Link:
Type:
Poster
Event:
Presenter's Name:
Tin Lay Nwe
Paper Code:
2132
Document Year:
2019
Cite

Document Files

icip2019-tinlay-v8.pdf

(10)

Subscribe

[1] Balaji Nataraj, Xie Shudong, Li Yiqun, Lin Dongyun, Dong Sheng , "DISCRIMINATIVE FEATURES FOR INCREMENTAL LEARNING CLASSIFIER", IEEE SigPort, 2019. [Online]. Available: http://sigport.org/4762. Accessed: Oct. 18, 2019.
@article{4762-19,
url = {http://sigport.org/4762},
author = {Balaji Nataraj; Xie Shudong; Li Yiqun; Lin Dongyun; Dong Sheng },
publisher = {IEEE SigPort},
title = {DISCRIMINATIVE FEATURES FOR INCREMENTAL LEARNING CLASSIFIER},
year = {2019} }
TY - EJOUR
T1 - DISCRIMINATIVE FEATURES FOR INCREMENTAL LEARNING CLASSIFIER
AU - Balaji Nataraj; Xie Shudong; Li Yiqun; Lin Dongyun; Dong Sheng
PY - 2019
PB - IEEE SigPort
UR - http://sigport.org/4762
ER -
Balaji Nataraj, Xie Shudong, Li Yiqun, Lin Dongyun, Dong Sheng . (2019). DISCRIMINATIVE FEATURES FOR INCREMENTAL LEARNING CLASSIFIER. IEEE SigPort. http://sigport.org/4762
Balaji Nataraj, Xie Shudong, Li Yiqun, Lin Dongyun, Dong Sheng , 2019. DISCRIMINATIVE FEATURES FOR INCREMENTAL LEARNING CLASSIFIER. Available at: http://sigport.org/4762.
Balaji Nataraj, Xie Shudong, Li Yiqun, Lin Dongyun, Dong Sheng . (2019). "DISCRIMINATIVE FEATURES FOR INCREMENTAL LEARNING CLASSIFIER." Web.
1. Balaji Nataraj, Xie Shudong, Li Yiqun, Lin Dongyun, Dong Sheng . DISCRIMINATIVE FEATURES FOR INCREMENTAL LEARNING CLASSIFIER [Internet]. IEEE SigPort; 2019. Available from : http://sigport.org/4762