Guided Self-Training based Semi-Supervised Learning for Fraud Detection

被引:3
|
作者
Kumar, Awanish [1 ]
Ghosh, Soumyadeep [1 ]
Verma, Janu [1 ]
机构
[1] Mastercard, AI Garage, Gurgaon, India
来源
3RD ACM INTERNATIONAL CONFERENCE ON AI IN FINANCE, ICAIF 2022 | 2022年
关键词
adversarial attack; vulnerability detection; vulnerability mitigation; transaction level vulnerability; black box vulnerability detection;
D O I
10.1145/3533271.3561783
中图分类号
F8 [财政、金融];
学科分类号
0202 ;
摘要
Semi supervised learning has attracted attention of AI researchers in the recent past, especially after the advent of deep learning methods and their success in several real world applications. Most deep learning models require large amounts of labelled data, which is expensive to obtain. Fraud detection is a very important problem for several industries and large amount of data is often available. However, obtaining labelled data is cumbersome and hence semi-supervised learning is perfectly positioned to aid us in building robust and accurate supervised models. In this work, we consider different kinds of fraud detection paradigms and show that a self-training based semi-supervised learning approach can produce significant improvements over a model that has been training on a limited set of labelled data. We propose a novel self-training approach by using a guided sharpening technique using a pair of autoencoders which provide useful cues for incorporating unlabelled data in the training process. We conduct thorough experiments on three different real world databases and analysis to showcase the effectiveness of the approach. On the elliptic bitcoin fraud dataset, we show that utilizing unlabelled data improves the F-1 score of the model trained on limited labelled data by around 10%.
引用
收藏
页码:148 / 155
页数:8
相关论文
共 7 条
  • [1] ASSBert: Active and semi-supervised bert for smart contract vulnerability detection
    Sun, Xiaobing
    Tu, Liangqiong
    Zhang, Jiale
    Cai, Jie
    Li, Bin
    Wang, Yu
    JOURNAL OF INFORMATION SECURITY AND APPLICATIONS, 2023, 73
  • [2] Exploring Public Data Vulnerabilities in Semi-Supervised Learning Models through Gray-box Adversarial Attack
    Jo, Junhyung
    Kim, Joongsu
    Suh, Young-Joo
    ELECTRONICS, 2024, 13 (05)
  • [3] Adversarial purification using a modified guided diffusion model and adaptive self-supervised training
    Zabihzadeh, Davood
    Alameer, Hawraa Razzaq Abed
    Mousavirad, Seyed Jalaleddin
    PHYSICA SCRIPTA, 2025, 100 (02)
  • [4] Union-adversarial Training for Deep Learning based on Surface Defect Detection
    Wu Q.
    Wen L.
    Jixie Gongcheng Xuebao/Journal of Mechanical Engineering, 2023, 59 (12): : 173 - 182
  • [5] SmiLe Net: A Supervised Graph Embedding-based Machine Learning Approach for NextG Vulnerability Detection
    Peng, Yifeng
    Yang, Jingda
    Arya, Sudhanshu
    Wang, Ying
    MILCOM 2023 - 2023 IEEE MILITARY COMMUNICATIONS CONFERENCE, 2023,
  • [6] ENSEMBLE ADVERSARIAL TRAINING BASED DEFENSE AGAINST ADVERSARIAL ATTACKS FOR MACHINE LEARNING-BASED INTRUSION DETECTION SYSTEM
    Haroon, M. S.
    Ali, H. M.
    NEURAL NETWORK WORLD, 2023, 33 (05) : 317 - 336
  • [7] SCVD-SA: A Smart Contract Vulnerability Detection Method based on Hybrid Deep Learning Model and Self-attention Mechanism
    Wang, Dongjie
    Chen, Jinfu
    Cai, Saihua
    Feng, Qiaowei
    Chen, Yuhao
    Hu, Xinyi
    2024 IEEE INTERNATIONAL CONFERENCE ON SOFTWARE ANALYSIS, EVOLUTION AND REENGINEERING-COMPANION, SANER-C 2024, 2024, : 207 - 214