Set-Based Boosting for Instance-Level Transfer on Multi-Classification

Haibo YIN; Jun-an YANG; Wei WANG; Hui LIU

doi:10.1587/transinf.2016EDP7373

Set-Based Boosting for Instance-Level Transfer on Multi-Classification

Haibo YIN, Jun-an YANG, Wei WANG, Hui LIU

Full Text Views

0

Share
Cite this

Summary :

Transfer boosting, a branch of instance-based transfer learning, is a commonly adopted transfer learning method. However, currently popular transfer boosting methods focus on binary classification problems even though there are many multi-classification tasks in practice. In this paper, we developed a new algorithm called MultiTransferBoost on the basis of TransferBoost for multi-classification. MultiTransferBoost firstly separated the multi-classification problem into several orthogonal binary classification problems. During each iteration, MultiTransferBoost boosted weighted instances from different source domains while each instance's weight was assigned and updated by evaluating the difficulty of the instance being correctly classified and the “transferability” of the instance's corresponding source domain to the target. The updating process repeated until it reached the predefined training error or iteration number. The weight update factors, which were analyzed and adjusted to minimize the Hamming loss of the output coding, strengthened the connections among the sub binary problems during each iteration. Experimental results demonstrated that MultiTransferBoost had better classification performance and less computational burden than existing instance-based algorithms using the One-Against-One (OAO) strategy.

Publication: IEICE TRANSACTIONS on Information Vol.E100-D No.5 pp.1079-1086

Publication Date: 2017/05/01

Publicized: 2017/01/26

Online ISSN: 1745-1361

DOI: 10.1587/transinf.2016EDP7373

Type of Manuscript: PAPER

Category: Pattern Recognition

Authors

Haibo YIN
  Electronic Engineering Institution
Jun-an YANG
  Electronic Engineering Institution
Wei WANG
  Electronic Engineering Institution
Hui LIU
  Electronic Engineering Institution

Keyword

transfer learning, boosting, multi-classification, Hamming loss, extreme learning machine

Cite this

Copy

Haibo YIN, Jun-an YANG, Wei WANG, Hui LIU, "Set-Based Boosting for Instance-Level Transfer on Multi-Classification" in IEICE TRANSACTIONS on Information, vol. E100-D, no. 5, pp. 1079-1086, May 2017, doi: 10.1587/transinf.2016EDP7373.
Abstract: Transfer boosting, a branch of instance-based transfer learning, is a commonly adopted transfer learning method. However, currently popular transfer boosting methods focus on binary classification problems even though there are many multi-classification tasks in practice. In this paper, we developed a new algorithm called MultiTransferBoost on the basis of TransferBoost for multi-classification. MultiTransferBoost firstly separated the multi-classification problem into several orthogonal binary classification problems. During each iteration, MultiTransferBoost boosted weighted instances from different source domains while each instance's weight was assigned and updated by evaluating the difficulty of the instance being correctly classified and the “transferability” of the instance's corresponding source domain to the target. The updating process repeated until it reached the predefined training error or iteration number. The weight update factors, which were analyzed and adjusted to minimize the Hamming loss of the output coding, strengthened the connections among the sub binary problems during each iteration. Experimental results demonstrated that MultiTransferBoost had better classification performance and less computational burden than existing instance-based algorithms using the One-Against-One (OAO) strategy.
URL: https://globals.ieice.org/en_transactions/information/10.1587/transinf.2016EDP7373/_p

Copy

@ARTICLE{e100-d_5_1079,
author={Haibo YIN, Jun-an YANG, Wei WANG, Hui LIU, },
journal={IEICE TRANSACTIONS on Information},
title={Set-Based Boosting for Instance-Level Transfer on Multi-Classification},
year={2017},
volume={E100-D},
number={5},
pages={1079-1086},
abstract={Transfer boosting, a branch of instance-based transfer learning, is a commonly adopted transfer learning method. However, currently popular transfer boosting methods focus on binary classification problems even though there are many multi-classification tasks in practice. In this paper, we developed a new algorithm called MultiTransferBoost on the basis of TransferBoost for multi-classification. MultiTransferBoost firstly separated the multi-classification problem into several orthogonal binary classification problems. During each iteration, MultiTransferBoost boosted weighted instances from different source domains while each instance's weight was assigned and updated by evaluating the difficulty of the instance being correctly classified and the “transferability” of the instance's corresponding source domain to the target. The updating process repeated until it reached the predefined training error or iteration number. The weight update factors, which were analyzed and adjusted to minimize the Hamming loss of the output coding, strengthened the connections among the sub binary problems during each iteration. Experimental results demonstrated that MultiTransferBoost had better classification performance and less computational burden than existing instance-based algorithms using the One-Against-One (OAO) strategy.},
keywords={},
doi={10.1587/transinf.2016EDP7373},
ISSN={1745-1361},
month={May},}

Copy

TY - JOUR
TI - Set-Based Boosting for Instance-Level Transfer on Multi-Classification
T2 - IEICE TRANSACTIONS on Information
SP - 1079
EP - 1086
AU - Haibo YIN
AU - Jun-an YANG
AU - Wei WANG
AU - Hui LIU
PY - 2017
DO - 10.1587/transinf.2016EDP7373
JO - IEICE TRANSACTIONS on Information
SN - 1745-1361
VL - E100-D
IS - 5
JA - IEICE TRANSACTIONS on Information
Y1 - May 2017
AB - Transfer boosting, a branch of instance-based transfer learning, is a commonly adopted transfer learning method. However, currently popular transfer boosting methods focus on binary classification problems even though there are many multi-classification tasks in practice. In this paper, we developed a new algorithm called MultiTransferBoost on the basis of TransferBoost for multi-classification. MultiTransferBoost firstly separated the multi-classification problem into several orthogonal binary classification problems. During each iteration, MultiTransferBoost boosted weighted instances from different source domains while each instance's weight was assigned and updated by evaluating the difficulty of the instance being correctly classified and the “transferability” of the instance's corresponding source domain to the target. The updating process repeated until it reached the predefined training error or iteration number. The weight update factors, which were analyzed and adjusted to minimize the Hamming loss of the output coding, strengthened the connections among the sub binary problems during each iteration. Experimental results demonstrated that MultiTransferBoost had better classification performance and less computational burden than existing instance-based algorithms using the One-Against-One (OAO) strategy.
ER -