개인신용평가 모형을 위한 딥러닝 활용에 대한 연구

• 영문명: A Study on the Application of Deep Learning for Credit Scoring System
• 발행기관: 한국시뮬레이션학회
• 저자명: 이군희(Gun Hee Lee) 유영범(Youngbeom Yoo) 하승인(SeungYin Ha)
• 간행물 정보: 『한국시뮬레이션학회 학술대회집』2017년 춘계학술대회 발표집, 4042~4044쪽, 전체 3쪽
• 주제분류: 공학 > 기타공학
• 파일형태: PDF
• 발행일자: 2017.04.26

영문 초록

In recent years;online P2P lending has also increased rapidly in Korea. P2P lending are a way for many small investors to make loans to people who need funds through an online platform without going through financial institutions;and the debtor repays part of the loan to investors every month. Recently;despite the spread of negative perceptions of P2P companies due to illegal loans of Lending Club. But P2P lending is alternative for person of low credit rating who do not received loans from financial institutions. Korean s P2P loan market is continuously growing;and the amount of loans per person has also increased due to the recent increase in corporate and small business lending in the micro lending for existing individuals. In this study;we use data from Lending Club to assess credit risk by loan characteristics. The evaluation method is a Convolutional Neural Network which shows high performance for image recognition and speech recognition among deep learning techniques. Most artificial neural network models have a problem of ignoring the shape of data as fully connected model. But Convolutional Neural Network complements the problems of other Neural network models by preserving the shape of the data. Therefore our research want to analyze the difference between the good and bad shape.

목차

1. 서론

2. 연구의 설계 및 측정방법

3. 실증 분석

4. 결론

키워드

deep learning Convolutional Neural Network credit scoring P2P lending