analiza aiqbench
CARTA introduces a self-supervised encoding method for financial transactions that solves the scalability issues of high-cardinality identifiers like merchant names without requiring expensive fraud labels. By leveraging heterogeneous tokenization and cross-attention, this approach enables efficient risk stratification and fraud probing on raw data, offering a practical alternative to fully supervised models in label-scarce environments.
CARTA: Self-Supervised Reusable Transaction Embeddings via Heterogeneous Tokenization and Cross-Attention
arXiv:2610.24920v1 Announce Type: new Abstract: Acquiring fraud labels for payment transactions is expensive and inconsistent across tasks, and high-cardinality identifiers such as merchant names make one-hot feature pipelines infeasible at scale. We present CARTA, a self-supervised encoder that learns a compact 16-dim embedding from raw transactions without labels; the frozen embedding is then reused for label-efficient fraud probing and exploratory unsupervised risk stratification without retraining the encoder. CARTA uses type-specific projections with learned embedding tables for high-cardinality identifiers, trained with token excision and InfoNCE contrastive loss. On IEEE-CIS, its linear probe at 1% of labels exceeds a raw-feature probe at 100%; on IBM TabFormer (2M rows), its frozen embedding reaches GBDT AUPRC 0.508, above frequency encoding (0.411) and the merchant-dropping baseline (0.022), though fully-supervised CatBoost is stronger. CARTA is a label-free, reusable representation that complements task-specific supervised models.
Abstract:Acquiring fraud labels for payment transactions is expensive and inconsistent across tasks, and high-cardinality identifiers such as merchant names make one-hot feature pipelines infeasible at scale. We present CARTA, a self-supervised encoder that learns a compact 16-dim embedding from raw transactions without labels; the frozen embedding is then reused for label-efficient fraud probing and exploratory unsupervised risk stratification without retraining the encoder. CARTA uses type-specific projections with learned embedding tables for high-cardinality identifiers, trained with token excision and InfoNCE contrastive loss. On IEEE-CIS, its linear probe at 1% of labels exceeds a raw-feature probe at 100%; on IBM TabFormer (2M rows), its frozen embedding reaches GBDT AUPRC 0.508, above frequency encoding (0.411) and the merchant-dropping baseline (0.022), though fully-supervised CatBoost is stronger. CARTA is a label-free, reusable representation that complements task-specific supervised models.
9 pages, 4 figures. Accepted at ICAIF '26 (7th ACM International Conference on AI in Finance), Milan, Italy, November 14-17, 2026
Machine Learning (cs.LG); Computation and Language (cs.CL)
arXiv:2610.24920 [cs.LG]
(or arXiv:2610.24920v1 [cs.LG] for this version)
https://doi.org/10.48550/arXiv.2610.24920 Focus to learn more arXiv-issued DOI via DataCite
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Text integral sub licența deschisă a publicației · 250 cuvinte