An Active Transfer Learning Framework for Protein-Protein Interaction Extraction

Lishuang LI; Xinyu HE; Jieqiong ZHENG; Degen HUANG; Fuji REN

doi:10.1587/transinf.2017EDP7232

An Active Transfer Learning Framework for Protein-Protein Interaction Extraction

Lishuang LI, Xinyu HE, Jieqiong ZHENG, Degen HUANG, Fuji REN

Full Text Views

0

Share
Cite this

Summary :

Protein-Protein Interaction Extraction (PPIE) from biomedical literatures is an important task in biomedical text mining and has achieved great success on public datasets. However, in real-world applications, the existing PPI extraction methods are limited to label effort. Therefore, transfer learning method is applied to reduce the cost of manual labeling. Current transfer learning methods suffer from negative transfer and lower performance. To tackle this problem, an improved TrAdaBoost algorithm is proposed, that is, relative distribution is introduced to initialize the weights of TrAdaBoost to overcome the negative transfer caused by domain differences. To make further improvement on the performance of transfer learning, an approach combining active learning with the improved TrAdaBoost is presented. The experimental results on publicly available PPI corpora show that our method outperforms TrAdaBoost and SVM when the labeled data is insufficient,and on document classification corpora, it also illustrates that the proposed approaches can achieve better performance than TrAdaBoost and TPTSVM in final, which verifies the effectiveness of our methods.

Publication: IEICE TRANSACTIONS on Information Vol.E101-D No.2 pp.504-511

Publication Date: 2018/02/01

Publicized: 2017/10/30

Online ISSN: 1745-1361

DOI: 10.1587/transinf.2017EDP7232

Type of Manuscript: PAPER

Category: Natural Language Processing

Authors

Lishuang LI
  Dalian University of Technology
Xinyu HE
  Dalian University of Technology
Jieqiong ZHENG
  Dalian University of Technology
Degen HUANG
  Dalian University of Technology
Fuji REN
  University of Tokushima

Keyword

protein-protein interaction, TrAdaBoost, actively transfer learning, relative distribution

Cite this

Copy

Lishuang LI, Xinyu HE, Jieqiong ZHENG, Degen HUANG, Fuji REN, "An Active Transfer Learning Framework for Protein-Protein Interaction Extraction" in IEICE TRANSACTIONS on Information, vol. E101-D, no. 2, pp. 504-511, February 2018, doi: 10.1587/transinf.2017EDP7232.
Abstract: Protein-Protein Interaction Extraction (PPIE) from biomedical literatures is an important task in biomedical text mining and has achieved great success on public datasets. However, in real-world applications, the existing PPI extraction methods are limited to label effort. Therefore, transfer learning method is applied to reduce the cost of manual labeling. Current transfer learning methods suffer from negative transfer and lower performance. To tackle this problem, an improved TrAdaBoost algorithm is proposed, that is, relative distribution is introduced to initialize the weights of TrAdaBoost to overcome the negative transfer caused by domain differences. To make further improvement on the performance of transfer learning, an approach combining active learning with the improved TrAdaBoost is presented. The experimental results on publicly available PPI corpora show that our method outperforms TrAdaBoost and SVM when the labeled data is insufficient,and on document classification corpora, it also illustrates that the proposed approaches can achieve better performance than TrAdaBoost and TPTSVM in final, which verifies the effectiveness of our methods.
URL: https://globals.ieice.org/en_transactions/information/10.1587/transinf.2017EDP7232/_p

Copy

@ARTICLE{e101-d_2_504,
author={Lishuang LI, Xinyu HE, Jieqiong ZHENG, Degen HUANG, Fuji REN, },
journal={IEICE TRANSACTIONS on Information},
title={An Active Transfer Learning Framework for Protein-Protein Interaction Extraction},
year={2018},
volume={E101-D},
number={2},
pages={504-511},
abstract={Protein-Protein Interaction Extraction (PPIE) from biomedical literatures is an important task in biomedical text mining and has achieved great success on public datasets. However, in real-world applications, the existing PPI extraction methods are limited to label effort. Therefore, transfer learning method is applied to reduce the cost of manual labeling. Current transfer learning methods suffer from negative transfer and lower performance. To tackle this problem, an improved TrAdaBoost algorithm is proposed, that is, relative distribution is introduced to initialize the weights of TrAdaBoost to overcome the negative transfer caused by domain differences. To make further improvement on the performance of transfer learning, an approach combining active learning with the improved TrAdaBoost is presented. The experimental results on publicly available PPI corpora show that our method outperforms TrAdaBoost and SVM when the labeled data is insufficient,and on document classification corpora, it also illustrates that the proposed approaches can achieve better performance than TrAdaBoost and TPTSVM in final, which verifies the effectiveness of our methods.},
keywords={},
doi={10.1587/transinf.2017EDP7232},
ISSN={1745-1361},
month={February},}

Copy

TY - JOUR
TI - An Active Transfer Learning Framework for Protein-Protein Interaction Extraction
T2 - IEICE TRANSACTIONS on Information
SP - 504
EP - 511
AU - Lishuang LI
AU - Xinyu HE
AU - Jieqiong ZHENG
AU - Degen HUANG
AU - Fuji REN
PY - 2018
DO - 10.1587/transinf.2017EDP7232
JO - IEICE TRANSACTIONS on Information
SN - 1745-1361
VL - E101-D
IS - 2
JA - IEICE TRANSACTIONS on Information
Y1 - February 2018
AB - Protein-Protein Interaction Extraction (PPIE) from biomedical literatures is an important task in biomedical text mining and has achieved great success on public datasets. However, in real-world applications, the existing PPI extraction methods are limited to label effort. Therefore, transfer learning method is applied to reduce the cost of manual labeling. Current transfer learning methods suffer from negative transfer and lower performance. To tackle this problem, an improved TrAdaBoost algorithm is proposed, that is, relative distribution is introduced to initialize the weights of TrAdaBoost to overcome the negative transfer caused by domain differences. To make further improvement on the performance of transfer learning, an approach combining active learning with the improved TrAdaBoost is presented. The experimental results on publicly available PPI corpora show that our method outperforms TrAdaBoost and SVM when the labeled data is insufficient,and on document classification corpora, it also illustrates that the proposed approaches can achieve better performance than TrAdaBoost and TPTSVM in final, which verifies the effectiveness of our methods.
ER -