Reproduce “InterFuse: Interaction-Guided Adaptive Fusion of Heterogeneous Models for Payment Risk Ranking” 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

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

InterFuse ranks payment risks by combining outputs from independently trained models, retaining each model’s prediction and input coverage. It lets model representations exchange case-specific information, then uses a shared fusion state to weight available models for each event. On IBM CCT, it reaches mean Recall@1% of 61.46%, versus UMI’s 53.37%. These results concern ranking under a limited review budget; upstream models remain frozen and only fusion is trained.

Summary by alphaXiv.

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