Incremental learning framework for real‐world fraud detection environment

Abstract

For detecting malicious bidding activities in e-auctions, this study develops a chunk-based incremental learning framework that can operate in real-world auction settings. The self-adaptive framework first classifies incoming bidder chunks to counter fraud in each auction and take necessary actions. The fraud classifier is then adjusted with confident bidders' labels validated via bidder verification and one-class classification. Based on real fraud data produced from commercial auctions, we conduct an extensive experimental study wherein the classifier is adapted incrementally using only relevant bidding data while evaluating the subsequent adjusted models' detection and misclassification rates. We also compare our classifier with static learning and learning without data relevancy. K E Y W O R D S chunk-based incremental learning, fraud detection, imbalanced data, incremental memory model, incremental SGD, one-class SVM

Incremental Learning with Self-labeling of Incoming High-dimensional Data

“…Lastly, to detect malicious bidders in e-auctions, the recent study [14] developed a selfadaptive chunk-based incremental learning framework. The self-adaptive framework can function in real-world auction settings, which first classifies incoming bidder chunks for countering the fraud bidders in every auction and takes required measures.…”

“…Incremental learning targets to improve an existing model gradually with new data, without retraining from the beginning, and without immediately forgetting the learned knowledge [3,4]. For this purpose, we should guarantee the stability and plasticity of the classifier [14]. Indeed, our incremental classifier should be stable enough to retain the information of the current and previous chunks a little longer in the memory and should forget past chunks gradually.…”

Isolation Forest Based on Minimal Spanning Tree

“…Solving the problem is a key factor affecting a large number of related issues. There are many applications in such fields as: Monitoring of intelligent control systems for fault diagnosis [1], [2], ensuring cybersecurity [3], detecting attacks on autonomous vehicle control systems [4], or repairing and deleting records in databases [5]. Another field of application worth mentioning is a wide group of medical tasks [6].…”

An efficient credit card fraud detection approach using cost‐sensitive weak learner with imbalanced dataset

“…Boosting and bagging are the most commonly used techniques in this group. A fraud detection approach has been used to detect anomalies and class imbalanced problems based on the sampling technique and machine learning technique 12‐14 . Still, it is an unsolved problem and an open research issue.…”

“…A fraud detection approach has been used to detect anomalies and class imbalanced problems based on the sampling technique and machine learning technique. [12][13][14] Still, it is an unsolved problem and an open research issue. The performance of machine learning techniques (MLTs) has decreased when an imbalanced credit card dataset.…”