Cost-sensitive parallel learning framework for insurance intelligence operation

Xinxin Jiang, Shirui Pan, Guodong Long, Fei Xiong, Jing Jiang, Chengqi Zhang

Research output: Contribution to journalArticleResearchpeer-review

Abstract

Recent advancements in artificial intelligence (AI) are providing the insurance industry with new opportunities to create tailored solutions and services based on newfound knowledge of consumers, and the execution of enhanced operations and business functions. However, insurance data is heterogeneous, and imbalanced class distribution with low frequency and high dimensions presents four major challenges to machine learning in real-world business. Traditional machine learning algorithms can typically only be applied to standard data sets, which are normally homogeneous and balanced. In this paper, we focus on an efficient cost-sensitive parallel learning framework (CPLF) to enhance insurance operations with a deep learning approach that does not require pre-processing. Our approach comprises a novel, unified, end-to-end cost-sensitive parallel neural network that learns real-world heterogeneous data. 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. We also study the CPLF-based architecture for a real-world insurance intelligence operation system, and demonstrate fraud detection experiments on this system. The results of comparative experiments on real-world insurance data sets reflecting actual business cases demonstrate the effectiveness of our design.

Original languageEnglish
Number of pages11
JournalIEEE Transactions on Industrial Electronics
DOIs
Publication statusAccepted/In press - 2019
Externally publishedYes

Keywords

  • deep learning
  • heterogeneous data
  • imbalanced data
  • insurance operation
  • neural network

Cite this

@article{f79f3057190841eaa0ca6074532d13a8,
title = "Cost-sensitive parallel learning framework for insurance intelligence operation",
abstract = "Recent advancements in artificial intelligence (AI) are providing the insurance industry with new opportunities to create tailored solutions and services based on newfound knowledge of consumers, and the execution of enhanced operations and business functions. However, insurance data is heterogeneous, and imbalanced class distribution with low frequency and high dimensions presents four major challenges to machine learning in real-world business. Traditional machine learning algorithms can typically only be applied to standard data sets, which are normally homogeneous and balanced. In this paper, we focus on an efficient cost-sensitive parallel learning framework (CPLF) to enhance insurance operations with a deep learning approach that does not require pre-processing. Our approach comprises a novel, unified, end-to-end cost-sensitive parallel neural network that learns real-world heterogeneous data. 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. We also study the CPLF-based architecture for a real-world insurance intelligence operation system, and demonstrate fraud detection experiments on this system. The results of comparative experiments on real-world insurance data sets reflecting actual business cases demonstrate the effectiveness of our design.",
keywords = "deep learning, heterogeneous data, imbalanced data, insurance operation, neural network",
author = "Xinxin Jiang and Shirui Pan and Guodong Long and Fei Xiong and Jing Jiang and Chengqi Zhang",
year = "2019",
doi = "10.1109/TIE.2018.2873526",
language = "English",
journal = "IEEE Transactions on Industrial Electronics",
issn = "0278-0046",
publisher = "IEEE, Institute of Electrical and Electronics Engineers",

}

Cost-sensitive parallel learning framework for insurance intelligence operation. / Jiang, Xinxin; Pan, Shirui; Long, Guodong; Xiong, Fei; Jiang, Jing; Zhang, Chengqi.

In: IEEE Transactions on Industrial Electronics, 2019.

Research output: Contribution to journalArticleResearchpeer-review

TY - JOUR

T1 - Cost-sensitive parallel learning framework for insurance intelligence operation

AU - Jiang, Xinxin

AU - Pan, Shirui

AU - Long, Guodong

AU - Xiong, Fei

AU - Jiang, Jing

AU - Zhang, Chengqi

PY - 2019

Y1 - 2019

N2 - Recent advancements in artificial intelligence (AI) are providing the insurance industry with new opportunities to create tailored solutions and services based on newfound knowledge of consumers, and the execution of enhanced operations and business functions. However, insurance data is heterogeneous, and imbalanced class distribution with low frequency and high dimensions presents four major challenges to machine learning in real-world business. Traditional machine learning algorithms can typically only be applied to standard data sets, which are normally homogeneous and balanced. In this paper, we focus on an efficient cost-sensitive parallel learning framework (CPLF) to enhance insurance operations with a deep learning approach that does not require pre-processing. Our approach comprises a novel, unified, end-to-end cost-sensitive parallel neural network that learns real-world heterogeneous data. 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. We also study the CPLF-based architecture for a real-world insurance intelligence operation system, and demonstrate fraud detection experiments on this system. The results of comparative experiments on real-world insurance data sets reflecting actual business cases demonstrate the effectiveness of our design.

AB - Recent advancements in artificial intelligence (AI) are providing the insurance industry with new opportunities to create tailored solutions and services based on newfound knowledge of consumers, and the execution of enhanced operations and business functions. However, insurance data is heterogeneous, and imbalanced class distribution with low frequency and high dimensions presents four major challenges to machine learning in real-world business. Traditional machine learning algorithms can typically only be applied to standard data sets, which are normally homogeneous and balanced. In this paper, we focus on an efficient cost-sensitive parallel learning framework (CPLF) to enhance insurance operations with a deep learning approach that does not require pre-processing. Our approach comprises a novel, unified, end-to-end cost-sensitive parallel neural network that learns real-world heterogeneous data. 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. We also study the CPLF-based architecture for a real-world insurance intelligence operation system, and demonstrate fraud detection experiments on this system. The results of comparative experiments on real-world insurance data sets reflecting actual business cases demonstrate the effectiveness of our design.

KW - deep learning

KW - heterogeneous data

KW - imbalanced data

KW - insurance operation

KW - neural network

UR - http://www.scopus.com/inward/record.url?scp=85054687688&partnerID=8YFLogxK

U2 - 10.1109/TIE.2018.2873526

DO - 10.1109/TIE.2018.2873526

M3 - Article

JO - IEEE Transactions on Industrial Electronics

JF - IEEE Transactions on Industrial Electronics

SN - 0278-0046

ER -