Reproduce “ID2Pattern: Pretraining Generalizable Identity Behavior Patterns for Transaction Risk Foundation Models” on OpenResearch

Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.

Universal app for Apple silicon and Intel. Requires macOS 11 or later.

About this paper

Ze Zhao, Haoyu Jiang, Weiyu Han, Ruobing Zuo, Xiao Tan, Shuai Chen, Tianyi Zhang

ID2Pattern encodes transaction identity reuse and recurrence across short, medium, and long histories as discrete patterns. On an industrial dataset of over 100 million sequences, it improves AUC-ROC by 4.13% relative to a semantic-attribute-only model; across six public datasets, average PR-AUC improvement is 8.83%. Relabeling identities while preserving relationships leaves PR-AUC unchanged, but does not demonstrate transfer to new populations.

Summary by alphaXiv.

Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.