불균형 정형 데이터를 위한 SMOTE와 변형 CycleGAN 기반 하이브리드 오버샘플링 기법 = A Hybrid Oversampling Technique for Imbalanced Structured Data based on SMOTE and Adapted CycleGAN

국문 초록 (Abstract)

이미지와 같은 비정형 데이터의 불균형 클래스 문제 해결에 있어 생산적 적대 신경망(generative adversarial network)에 기반한 오버샘플링 기법의 우수성이 알려짐에 따라 다양한 연구들이 이를 정형데이터의 불균형 문제 해결에도 적용하기 시작하였다. 그러나 이러한 연구들은 데이터의 형태를 비정형데이터 구조로 변경함으로써 정형 데이터의 특징을 정확하게 반영하지 못한다는 점이 문제로 지적되고있다. 본 연구에서는 이를 해결하기 위해 순환 생산적 적대 신경망(cycle GAN)을 정형 데이터의 구조에맞게 재구성하고 이를 SMOTE(synthetic minority oversampling technique) 기법과 결합한 하이브리드오버샘플링 기법을 제안하였다. 특히 기존 연구와 달리 생산적 적대 신경망을 구성함에 있어 1차원합성곱 신경망(1D-convolutional neural network)을 사용함으로써 기존 연구의 한계를 극복하고자 하였다.
본 연구에서 제안한 기법의 성능 비교를 위해 불균형 정형 데이터를 기반으로 오버샘플링을 진행하고그 결과를 SMOTE, ADASYN(adaptive synthetic sampling) 등과 같은 기존 기법과 비교하였다. 비교결과 차원이 많을수록, 불균형 정도가 심할수록 제안된 모형이 우수한 성능을 보이는 것으로 나타났다.
본 연구는 기존 연구와 달리 정형 데이터의 구조를 유지하면서 소수 클래스의 특징을 반영한오버샘플링을 통해 분류의 성능을 향상시켰다는 점에서 의의가 있다.

참고문헌 (Reference)

1 김예원 ; 유예림 ; 최홍용, "생성적 적대 신경망과 딥러닝을 활용한 이상거래탐지 시스템 모형" 한국경영정보학회 22 (22): 59-72, 2020

2 Gangwar, A. K, "WiP: Generative adversarial network for oversampling data in credit card fraud detection" 123-134, 2019

3 Arjovsky, M, "Wasserstein generative adversarial networks" 214-223, 2017

4 Statista, "Volume of data/information created, captured, copied, and consumed worldwide from 2010 to 2025"

5 Fiore, U, "Using generative adversarial networks for improving classification effectiveness in credit card fraud detection" 479 : 448-455, 2019

6 Radford, A, "Unsupervised representation learning with deep convolutional generative adversarial networks" 2016

7 Wang, J, "Unrolled GAN-based oversampling of credit card dataset for fraud detection" 858-861, 2022

8 Zhu, J. Y, "Unpaired image-to-image translation using cycle-consistent adversarial networks" 2223-2232, 2017

9 Tomek, I, "Two modifications of CNN" 6 (6): 769-772, 1976

10 Quintana, M, "Towards class-balancing human comfort datasets with GANs" 391-392, 2019

11 Xu, L, "Synthesizing Tabular Data using Conditional GAN" Massachusetts Institute of Technology 2020

12 Johnson, J. M, "Survey on deep learning with class imbalance" 6 : 27-, 2019

13 Refinitive, "Smarter humans. Smarter machines"

14 Sharma, A, "SMOTified-GAN for class imbalanced pattern classification problems" 10 : 30655-30665, 2022

15 Chawla, N. V, "SMOTE: Synthetic minority over-sampling technique" 16 : 321-357, 2002

16 Sáez, J. A, "SMOTE-IPF: Addressing the noisy and borderline examples problem in imbalanced classification by a re-sampling method with filtering" 291 : 184-203, 2015

17 Fangyu, W, "Research on imbalanced data set preprocessing based on deep learning" 75-79, 2021

18 Soltanzadeh, P, "RCSMOTE: Range-controlled synthetic minority over-sampling technique for handling the class imbalance problem" 542 : 92-111, 2020

19 Tek, F. B, "Parasite detection and identification for automated thin blood film malaria diagnosis" 114 : 21-32, 2010

20 Nazari, E, "On oversampling via generative adversarial networks under different data difficulty factors" 154 : 76-89, 2021

21 Gazzah, S, "New oversampling approaches based on polynomial fitting for imbalanced data sets" 677-684, 2008

22 Yang, Y, "Network intrusion detection based on supervised adversarial variational auto-encoder with regularization" 8 : 42169-42184, 2020

23 Mohammed, R, "Machine learning with oversampling and undersampling techniques: Overview study and experimental results" 243-248, 2020

24 Krawczyk, B, "Learning from imbalanced data: Open challenges and future directions" 5 (5): 221-232, 2016

25 He, H, "Learning from imbalanced data" 21 (21): 1263-1284, 2009

26 Islam, A, "KNNOR: An oversampling technique for imbalanced datasets" 115 : 108288-, 2022

27 IBM, "Inforgraphic-Extracting business value form the 4Vs of big data"

28 Ba, H, "Improving detection of credit card fraudulent transactions using generative adversarial networks"

29 Liu, Y, "Imbalanced text classification: A term weighting approach" 36 : 690-701, 2009

30 Krizhevsky, A, "ImageNet classification with deep convolutional neural networks" 60 (60): 84-90, 2017

31 Wise, J, "How much data is created every day in 2022?"

32 Kingma, D. P, "Glow: Generative flow with invertible 1x1 convolutions" 31 : 2018

33 Zhu, J.-Y, "Generative visual manipulation on the natural image manifold" 597-613, 2016

34 Wang, Z, "Generative adversarial networks: A survey and taxonomy" 54 (54): 1-38, 2022

35 Saxena, D, "Generative adversarial networks (GANs) challenges, solutions, and future directions" 54 (54): 1-42, 2022

36 Goodfellow, I. J, "Generative adversarial nets" 27 : 2672-2680, 2014

37 Mullick, S. S, "Generative adversarial minority oversampling" 1695-1704, 2019

38 Kate, P, "FinGAN: Generative adversarial network for analytical customer relationship management in banking and insurance"

39 Sambasivan, N, "Everyone wants to do the model work, not the data work”: Data cascades in high-stakes AI" 1-15, 2021

40 Khoshgoftaar, T. M, "Ensemble vs. data sampling: Which option is best suited to improve classification performance of imbalanced bioinformatics data?" 705-712, 2015

41 Douzas, G, "Effective data generation for imbalanced learning using conditional generative adversarial networks" 91 : 464-471, 2018

42 Chawla, N. V, "Editorial: Special issue on learning from imbalanced data sets" 6 (6): 1-6, 2004

43 Fernández-Delgado, M, "Do we need hundreds of classifiers to solve real world classification problems?" 15 (15): 3133-3181, 2014

44 Zhou, F, "Deep learning fault diagnosis method based on global optimization GAN for unbalanced data" 187 : 104837-, 2020

45 Ling, C. X, "Data mining for direct marketing: Problems and solutions" 73-79, 1998

46 Chen, H, "Data evaluation and enhancement for quality improvement of machine learning" 70 (70): 831-847, 2021

47 Dlamini, G, "DGM: A data generative model to improve minority class presence in anomaly detection domain" 33 (33): 13635-13646, 2021

48 최형욱 ; 이승현 ; 김형훈 ; 서용철, "CycleGAN을 활용한 항공영상 학습 데이터 셋 보완 기법에 관한 연구" 한국측량학회 38 (38): 499-509, 2020

49 Pathak, D, "Context encoders: Feature learning by inpainting" 2536-2544, 2016

50 Mirza, M, "Conditional generative adversarial nets"

51 Engelmann, J, "Conditional Wasserstein GAN-based oversampling of tabular data for imbalanced learning" 174 : 114582-, 2021

52 Han, H, "Borderline-SMOTE: A new over-sampling method in imbalanced data sets learning" 3644 (3644): 878-887, 2005

53 Wilson, D. L, "Asymptotic properties of nearest neighbor rules using edited data" 2 (2): 408-421, 1972

54 Cao, Q, "Applying over-sampling technique based on data density and cost-sensitive SVM to imbalanced learning" 543-548, 2011

55 Chandola, V, "Anomaly detection: A survey" 41 (41): 1-58, 2009

56 Fernández, A, "An insight into imbalanced big data classification: Outcomes and challenges" 3 : 105-120, 2017

57 Thejas G. S, "An extension of synthetic minority oversampling technique based on Kalman filter for imbalanced datasets" 8 : 100267-, 2022

58 Bai, S, "An empirical evaluation of generic convolutional and recurrent networks for sequence modeling"

59 Kovács, G, "An empirical comparison and evaluation of minority oversampling techniques on a large number of imbalanced datasets" 83 : 105662-, 2019

60 Yap, B. W, "An application of oversampling, undersampling, bagging and boosting in handling imbalanced datasets" 13-22, 2013

61 Deepa, T, "An E-SMOTE technique for feature selection in high-dimensional imbalanced dataset" 2 : 322-324, 2011

62 He, H, "ADASYN: Adaptive synthetic sampling approach for imbalanced learning" 1322-1328, 2008

63 Bosu, M. F, "A taxonomy of data quality challenges in empirical software engineering" 97-106, 2013

64 Leevy, J. L, "A survey on addressing high-class imbalance in big data" 5 : 42-, 2018

65 Gui, J, "A review on generative adversarial networks: Algorithms, theory, and applications"

66 Zhou, B, "A quasi-linear SVM combined with assembled SMOTE for imbalanced data classification" 1-7, 2013

67 Aydilek, I. B, "A hybrid method for imputation of missing values using optimized fuzzy c-means with support vector regression and a genetic algorithm" 233 : 25-35, 2013

68 Silver, D, "A general reinforcement learning algorithm that masters chess, shogi, and Go through self-play" 362 (362): 1140-1144, 2018

69 Zhu, B, "A GAN-based hybrid sampling method for imbalanced customer classification" 609 : 1397-1411, 2022