Cost-sensitive hybrid neural networks for heterogeneous and imbalanced data

Xinxin Jiang, Shirui Pan, Guodong Long, Jiang Chang, Jing Jiang, Chengqi Zhang

Research output: Chapter in Book/Report/Conference proceedingConference PaperResearchpeer-review

Abstract

Analyzing accumulated data has recently attracted huge attention for its ability to generate values by identifying useful information and providing an edge in global business competition. However, heterogeneous data and imbalanced class distribution present two major challenges to machine learning with real-world business data. Traditional machine learning algorithms can typically only be applied to standard data sets, which are normally homogeneous and balanced. These algorithms narrow complex data into a homogeneous, a balanced data space an inefficient process that requires a significant amount of pre-processing. In this paper, we focus on an efficient solution to the challenges with heterogeneous and imbalanced data sets that does not require pre-processing. Our approach comprises a novel, unified, end-to-end cost-sensitive hybrid neural network that learns real-world heterogeneous data via a parallel network architecture. A specifically-designed cost-sensitive matrix then automatically generates a robust model for learning minority classifications. And the parameters of both the cost-sensitive matrix and the hybrid neural network are alternately but jointly optimized during training. The results of comparative experiments on six real-world data sets reflecting actual business cases, including insurance fraud detection and mobile customer demographics, indicate that the proposed approach demonstrates superior performance over baseline procedures.

Original languageEnglish
Title of host publication2018 International Joint Conference on Neural Networks (IJCNN) - 2012 Proceedings
EditorsLeandro Minku, Rodrigo Soares
Place of PublicationPiscataway NJ USA
PublisherIEEE, Institute of Electrical and Electronics Engineers
Pages3446-3453
Number of pages8
ISBN (Electronic)9781509060146
DOIs
Publication statusPublished - 2018
Externally publishedYes
EventIEEE International Joint Conference on Neural Networks 2018 - Rio de Janeiro, Brazil
Duration: 8 Jul 201813 Jul 2018
Conference number: 2018
http://www.ecomp.poli.br/~wcci2018/

Conference

ConferenceIEEE International Joint Conference on Neural Networks 2018
Abbreviated titleIJCNN 2018
CountryBrazil
CityRio de Janeiro
Period8/07/1813/07/18
Internet address

Keywords

  • heterogeneous
  • hybrid neural network
  • imbalanced data

Cite this

Jiang, X., Pan, S., Long, G., Chang, J., Jiang, J., & Zhang, C. (2018). Cost-sensitive hybrid neural networks for heterogeneous and imbalanced data. In L. Minku, & R. Soares (Eds.), 2018 International Joint Conference on Neural Networks (IJCNN) - 2012 Proceedings (pp. 3446-3453). [8489420] Piscataway NJ USA: IEEE, Institute of Electrical and Electronics Engineers. https://doi.org/10.1109/IJCNN.2018.8489420
Jiang, Xinxin ; Pan, Shirui ; Long, Guodong ; Chang, Jiang ; Jiang, Jing ; Zhang, Chengqi. / Cost-sensitive hybrid neural networks for heterogeneous and imbalanced data. 2018 International Joint Conference on Neural Networks (IJCNN) - 2012 Proceedings. editor / Leandro Minku ; Rodrigo Soares. Piscataway NJ USA : IEEE, Institute of Electrical and Electronics Engineers, 2018. pp. 3446-3453
@inproceedings{5da492975953481c8bce69843ce92691,
title = "Cost-sensitive hybrid neural networks for heterogeneous and imbalanced data",
abstract = "Analyzing accumulated data has recently attracted huge attention for its ability to generate values by identifying useful information and providing an edge in global business competition. However, heterogeneous data and imbalanced class distribution present two major challenges to machine learning with real-world business data. Traditional machine learning algorithms can typically only be applied to standard data sets, which are normally homogeneous and balanced. These algorithms narrow complex data into a homogeneous, a balanced data space an inefficient process that requires a significant amount of pre-processing. In this paper, we focus on an efficient solution to the challenges with heterogeneous and imbalanced data sets that does not require pre-processing. Our approach comprises a novel, unified, end-to-end cost-sensitive hybrid neural network that learns real-world heterogeneous data via a parallel network architecture. A specifically-designed cost-sensitive matrix then automatically generates a robust model for learning minority classifications. And the parameters of both the cost-sensitive matrix and the hybrid neural network are alternately but jointly optimized during training. The results of comparative experiments on six real-world data sets reflecting actual business cases, including insurance fraud detection and mobile customer demographics, indicate that the proposed approach demonstrates superior performance over baseline procedures.",
keywords = "heterogeneous, hybrid neural network, imbalanced data",
author = "Xinxin Jiang and Shirui Pan and Guodong Long and Jiang Chang and Jing Jiang and Chengqi Zhang",
year = "2018",
doi = "10.1109/IJCNN.2018.8489420",
language = "English",
pages = "3446--3453",
editor = "Leandro Minku and Rodrigo Soares",
booktitle = "2018 International Joint Conference on Neural Networks (IJCNN) - 2012 Proceedings",
publisher = "IEEE, Institute of Electrical and Electronics Engineers",
address = "United States of America",

}

Jiang, X, Pan, S, Long, G, Chang, J, Jiang, J & Zhang, C 2018, Cost-sensitive hybrid neural networks for heterogeneous and imbalanced data. in L Minku & R Soares (eds), 2018 International Joint Conference on Neural Networks (IJCNN) - 2012 Proceedings., 8489420, IEEE, Institute of Electrical and Electronics Engineers, Piscataway NJ USA, pp. 3446-3453, IEEE International Joint Conference on Neural Networks 2018, Rio de Janeiro, Brazil, 8/07/18. https://doi.org/10.1109/IJCNN.2018.8489420

Cost-sensitive hybrid neural networks for heterogeneous and imbalanced data. / Jiang, Xinxin; Pan, Shirui; Long, Guodong; Chang, Jiang; Jiang, Jing; Zhang, Chengqi.

2018 International Joint Conference on Neural Networks (IJCNN) - 2012 Proceedings. ed. / Leandro Minku; Rodrigo Soares. Piscataway NJ USA : IEEE, Institute of Electrical and Electronics Engineers, 2018. p. 3446-3453 8489420.

Research output: Chapter in Book/Report/Conference proceedingConference PaperResearchpeer-review

TY - GEN

T1 - Cost-sensitive hybrid neural networks for heterogeneous and imbalanced data

AU - Jiang, Xinxin

AU - Pan, Shirui

AU - Long, Guodong

AU - Chang, Jiang

AU - Jiang, Jing

AU - Zhang, Chengqi

PY - 2018

Y1 - 2018

N2 - Analyzing accumulated data has recently attracted huge attention for its ability to generate values by identifying useful information and providing an edge in global business competition. However, heterogeneous data and imbalanced class distribution present two major challenges to machine learning with real-world business data. Traditional machine learning algorithms can typically only be applied to standard data sets, which are normally homogeneous and balanced. These algorithms narrow complex data into a homogeneous, a balanced data space an inefficient process that requires a significant amount of pre-processing. In this paper, we focus on an efficient solution to the challenges with heterogeneous and imbalanced data sets that does not require pre-processing. Our approach comprises a novel, unified, end-to-end cost-sensitive hybrid neural network that learns real-world heterogeneous data via a parallel network architecture. A specifically-designed cost-sensitive matrix then automatically generates a robust model for learning minority classifications. And the parameters of both the cost-sensitive matrix and the hybrid neural network are alternately but jointly optimized during training. The results of comparative experiments on six real-world data sets reflecting actual business cases, including insurance fraud detection and mobile customer demographics, indicate that the proposed approach demonstrates superior performance over baseline procedures.

AB - Analyzing accumulated data has recently attracted huge attention for its ability to generate values by identifying useful information and providing an edge in global business competition. However, heterogeneous data and imbalanced class distribution present two major challenges to machine learning with real-world business data. Traditional machine learning algorithms can typically only be applied to standard data sets, which are normally homogeneous and balanced. These algorithms narrow complex data into a homogeneous, a balanced data space an inefficient process that requires a significant amount of pre-processing. In this paper, we focus on an efficient solution to the challenges with heterogeneous and imbalanced data sets that does not require pre-processing. Our approach comprises a novel, unified, end-to-end cost-sensitive hybrid neural network that learns real-world heterogeneous data via a parallel network architecture. A specifically-designed cost-sensitive matrix then automatically generates a robust model for learning minority classifications. And the parameters of both the cost-sensitive matrix and the hybrid neural network are alternately but jointly optimized during training. The results of comparative experiments on six real-world data sets reflecting actual business cases, including insurance fraud detection and mobile customer demographics, indicate that the proposed approach demonstrates superior performance over baseline procedures.

KW - heterogeneous

KW - hybrid neural network

KW - imbalanced data

UR - http://www.scopus.com/inward/record.url?scp=85056519128&partnerID=8YFLogxK

U2 - 10.1109/IJCNN.2018.8489420

DO - 10.1109/IJCNN.2018.8489420

M3 - Conference Paper

SP - 3446

EP - 3453

BT - 2018 International Joint Conference on Neural Networks (IJCNN) - 2012 Proceedings

A2 - Minku, Leandro

A2 - Soares, Rodrigo

PB - IEEE, Institute of Electrical and Electronics Engineers

CY - Piscataway NJ USA

ER -

Jiang X, Pan S, Long G, Chang J, Jiang J, Zhang C. Cost-sensitive hybrid neural networks for heterogeneous and imbalanced data. In Minku L, Soares R, editors, 2018 International Joint Conference on Neural Networks (IJCNN) - 2012 Proceedings. Piscataway NJ USA: IEEE, Institute of Electrical and Electronics Engineers. 2018. p. 3446-3453. 8489420 https://doi.org/10.1109/IJCNN.2018.8489420