cs.CRMar 13, 2026

FraudFox: Adaptable Fraud Detection in the Real World

Authors: Matthew Butler, Yi Fan, Christos Faloutsos

Organizations: Amazon.com, Inc., 410 Terry Ave N, Seattle WA 98109, U.S.A. · Carnegie Mellon University, 5000 Forbes Avenue, Pittsburgh, U.S.A.

Read PDF
Ask AI
Cite

Copy citation

DOINot available for this paper

Abstract

The proposed method (FraudFox) provides solutions to adversarial attacks in a resource constrained environment. We focus on questions like the following: How suspicious is Smith', trying to buy \$500 shoes, on Monday 3am? How to merge the risk scores, from a handful of risk-assessment modules (oracles') in an adversarial environment? More importantly, given historical data (orders, prices, and what-happened afterwards), and business goals/restrictions, which transactions, like the Smith' transaction above, which ones should we pass', versus send to human investigators? The business restrictions could be: at most $x$ investigations are feasible', or at most $yy lost due to fraud'. These are the two research problems we focus on, in this work. One approach to address the first problem (`oracle-weighting'), is by using Extended Kalman Filters with dynamic importance weights, to automatically and continuously update our weights for each 'oracle'. For the second problem, we show how to derive an optimal decision surface, and how to compute the Pareto optimal set, to allow what-if questions. An important consideration is adaptation: Fraudsters will change their behavior, according to our past decisions; thus, we need to adapt accordingly. The resulting system, \method, is scalable, adaptable to changing fraudster behavior, effective, and already in production at Amazon. FraudFox augments a fraud prevention sub-system and has led to significant performance gains.

Figures & tables

Figure 1: Business Impact of FraudFox : Visible loss reduction when applying it to real word data. Losses vs. time - 1st without dynamically blending a subset of fraud indicators (blue), i.e., without FraudFox ; 2nd with dynamically blending (orange), i.e. with FraudFox introduced at the dashed-black time tick.
PropertyL++L^{++} Muhlbaier and Polikar (2007)AC Dalvi et al. (2004)DLM Penny and Roberts (1999)FraudFox
Auto-adapts to adversaries ✔ ✔
Auto-adapts to non-stationary✔ ✔✔
Auto-adapts to business rec’s   ✔
Effective/Deployed   ✔
Uses ensemble✔ ✔✔
Table 1: FraudFox matches all specs , while competitors miss one or more of the features.
SymbolsDefinitions
kkcount of Oracles
w⃗t\vec{w}_{t}weights of the Oracles (k x 1 vector) at time tt
Σt\mathbf{\Sigma}_{t}covariance matrix of the Oracles (k x k matrix)
Kt\mathbf{K}_{t}Kalman Gain
Ht\mathbf{H}_{t}Observation matrix (Eq 6 )
β\betahyper-parameter of weighting policy (Eq 4 )
Table 2: Symbols and definitions
Figure 2: Illustration of Decision Surface and Adversarial Behavior. In the (fraud-score vs. order value) plot, when we fix the decision surface (blue curve), fraudulent orders (red crosses) will probably flock just below it. FraudFox -I is blocking exactly this behavior.
Figure 3: Cost Benefit Analysis. Profit, for each of the four cases.
Figure 4: Business Adaptation and Pareto Front. FraudFox precomputes the Pareto front, in preparation of changes in business requirements. Scatter plot of model performance in business metric space (# of investigations, vs $ fraud captured per investigation). Every point is a possible parameter choice of FraudFox . Points with empty light blue NE quadrants are non-dominating and thus form the Pareto front (green ’x’ points).
Figure 5: FraudFox Adapts : Synthetic Example. As expected, both our versions (equal-weights, in ’green’ and exponentially-decaying weights, in ’red’) responds to a change/shock at tt =30, and eventually recover from the shock. Without adaptation (’blue’), the behavior suffers.
More in this topic

Financial Fraud Detection

Financial fraud detection identifies suspicious transactions, accounts, claims, and behavior to prevent losses and guide investigation.

See all 135 papers

Explore similar work

CardsList

  1. SR-Fraud: An Outcome-Supervised Reflective LLM Agent Framework for Non-Stationary Payment Fraud Detection

    Sep 23, 2026Xuwei Tan, Yao Ma, Xueru ZhangConcept DriftFinancial Fraud Detection

    Coinbase, Inc. · The Ohio State University

  2. Toward Auditable Fraud Detection: Combining Graph Features, Model Explanations, and Agentic Case Investigation

    Jul 21, 2026Rahil SharmaAlgorithmic AuditingTabular Anomaly Detection

    Vrije Universiteit Amsterdam

  3. FraudBench: Protocol-Sensitive Benchmarking of Adversarial Robustness for Financial Risk Assessment

    Aug 25, 2026Xitong Zeng, Zhaoge Bi, Yitian Yang +2Attention-Based Adversarial AttacksAdversarial Robustness

    School of Electrical and Computer Engineering The University of Sydney Sydney, Australia · School of Computing Macquarie University Sydney, Australia