TREASURE: The Visa Payment Foundation Model for High-Volume Transaction Understanding
Summary: arXiv:2511.19693v3 Announce Type: replace-cross
Abstract: Payment networks form the backbone of modern commerce, generating high volumes of transaction records from daily activities. Properly modeling this data can enable applications such as abnormal behavior detection and consumer-level insights for hyper-personalized experiences, ultimately improving people’s lives. In this paper, we present TREASURE, TRansformer Engine As Scalable Universal transaction Representation Encoder, a multipurpose transformer-based foundation model specifically designed for transaction data.
The model simultaneously captures both consumer behavior and payment network signals (such as response codes and system flags), providing comprehensive information necessary for applications like accurate recommendation systems and abnormal behavior detection. Verified with industry-grade datasets, TREASURE features three key capabilities:
- Input Module: TREASURE includes dedicated sub-modules for static and dynamic attributes, enabling more efficient training and inference.
- Training Paradigm: The model employs an efficient and effective training paradigm for predicting high-cardinality categorical attributes.
- Performance Enhancement: TREASURE demonstrates effectiveness as both a standalone model that increases abnormal behavior detection performance by 111% over production systems and as an embedding provider that enhances recommendation models by 104%.
Through extensive ablation studies, benchmarks against production models, and case studies, we present key insights that highlight the valuable knowledge gained from developing TREASURE.
Key Features of TREASURE
TREASURE’s architecture is designed to address the specific challenges posed by transaction data in payment networks. Below are some of the notable features that set TREASURE apart from existing models:
- Multipurpose Design: Unlike traditional models that focus on either consumer behavior or payment signals, TREASURE integrates both aspects to provide a holistic view of transaction dynamics.
- Scalability: The transformer-based framework allows TREASURE to scale effectively, handling vast amounts of transaction data without a loss in performance.
- Real-Time Insights: With its efficient processing capabilities, TREASURE can generate real-time insights, enabling businesses to respond swiftly to emerging trends and potential fraud.
Conclusion
The introduction of TREASURE marks a significant advancement in the modeling of transaction data within payment networks. Its ability to enhance abnormal behavior detection and improve recommendation systems not only benefits businesses but also enriches consumer experiences. As payment networks continue to evolve, TREASURE is poised to play a crucial role in shaping the future of transactional analytics.
For further details, please refer to the full paper on arXiv:2511.19693v3.