Efficient Class-Incremental Learning Based on Bag-of-Sequencelets Model for Activity Recognition
Summary :
We propose a class-incremental learning framework for human activity recognition based on the Bag-of-Sequencelets model (BoS). The framework updates learned models efficiently without having to relearn them when training data of new classes are added. In this framework, all types of features including hand-crafted features and Convolutional Neural Networks (CNNs) based features and combinations of those features can be used as features for videos. Compared with the original BoS, the new framework can reduce the learning time greatly with little loss of classification accuracy.
- Publication
- IEICE TRANSACTIONS on Fundamentals Vol.E102-A No.9 pp.1293-1302
- Publication Date
- 2019/09/01
- Publicized
- Online ISSN
- 1745-1337
- DOI
- 10.1587/transfun.E102.A.1293
- Type of Manuscript
- PAPER
- Category
- Vision
Authors
Jong-Woo LEE
Pohang University of Science and Technology (POSTECH)
Ki-Sang HONG
Pohang University of Science and Technology (POSTECH)
Keyword
Latest Issue
Copyrights notice of machine-translated contents
The copyright of the original papers published on this site belongs to IEICE. Unauthorized use of the original or translated papers is prohibited. See IEICE Provisions on Copyright for details.
Cite this
Copy
Jong-Woo LEE, Ki-Sang HONG, "Efficient Class-Incremental Learning Based on Bag-of-Sequencelets Model for Activity Recognition" in IEICE TRANSACTIONS on Fundamentals,
vol. E102-A, no. 9, pp. 1293-1302, September 2019, doi: 10.1587/transfun.E102.A.1293.
Abstract: We propose a class-incremental learning framework for human activity recognition based on the Bag-of-Sequencelets model (BoS). The framework updates learned models efficiently without having to relearn them when training data of new classes are added. In this framework, all types of features including hand-crafted features and Convolutional Neural Networks (CNNs) based features and combinations of those features can be used as features for videos. Compared with the original BoS, the new framework can reduce the learning time greatly with little loss of classification accuracy.
URL: https://global.ieice.org/en_transactions/fundamentals/10.1587/transfun.E102.A.1293/_p
Copy
@ARTICLE{e102-a_9_1293,
author={Jong-Woo LEE, Ki-Sang HONG, },
journal={IEICE TRANSACTIONS on Fundamentals},
title={Efficient Class-Incremental Learning Based on Bag-of-Sequencelets Model for Activity Recognition},
year={2019},
volume={E102-A},
number={9},
pages={1293-1302},
abstract={We propose a class-incremental learning framework for human activity recognition based on the Bag-of-Sequencelets model (BoS). The framework updates learned models efficiently without having to relearn them when training data of new classes are added. In this framework, all types of features including hand-crafted features and Convolutional Neural Networks (CNNs) based features and combinations of those features can be used as features for videos. Compared with the original BoS, the new framework can reduce the learning time greatly with little loss of classification accuracy.},
keywords={},
doi={10.1587/transfun.E102.A.1293},
ISSN={1745-1337},
month={September},}
Copy
TY - JOUR
TI - Efficient Class-Incremental Learning Based on Bag-of-Sequencelets Model for Activity Recognition
T2 - IEICE TRANSACTIONS on Fundamentals
SP - 1293
EP - 1302
AU - Jong-Woo LEE
AU - Ki-Sang HONG
PY - 2019
DO - 10.1587/transfun.E102.A.1293
JO - IEICE TRANSACTIONS on Fundamentals
SN - 1745-1337
VL - E102-A
IS - 9
JA - IEICE TRANSACTIONS on Fundamentals
Y1 - September 2019
AB - We propose a class-incremental learning framework for human activity recognition based on the Bag-of-Sequencelets model (BoS). The framework updates learned models efficiently without having to relearn them when training data of new classes are added. In this framework, all types of features including hand-crafted features and Convolutional Neural Networks (CNNs) based features and combinations of those features can be used as features for videos. Compared with the original BoS, the new framework can reduce the learning time greatly with little loss of classification accuracy.
ER -
English
