Explainable artificial intelligence models using SHAP enhanced CatBoost, Bi-GRU with attention, and Tab Transformer
Markapurapu John Dana Ebinezer, Bondalapu Chaitanya Krishna · 2026
Read Explainable artificial intelligence models using SHAP enhanced CatBoost, Bi-GRU with attention, and Tab Transformer by Markapurapu John Dana Ebinezer, Bondalapu Chaitanya Krishna (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/pmc-13332054.txt.
Description
Insurance fraud is hard to catch because claims data are lopsided and messy, with many kinds of information mixed together. The authors compare three model families under one test protocol: CatBoost, a tree-based method; a bidirectional recurrent network with attention; and TabTransformer, built for categorical data. The dataset holds 4,000 life insurance claims described by 83 attributes. All models go through the same cleaning, scaling, and feature selection steps before training.
The study's new element is a shared interpretability framework that pairs SHAP feature attributions with attention analysis, so the models' reasoning can be compared directly. CatBoost proves most precise on honest claims, the recurrent network recalls the most fraud, and TabTransformer balances accuracy, clarity, and cost best. Claim amount, policy tenure, and diagnosis stand out in both explanation methods. The authors offer the setup as a dependable benchmark for fraud detection research.
Catalog record
| CREATOR | Markapurapu John Dana Ebinezer, Bondalapu Chaitanya Krishna |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 81 KB |
| VERSION | 1.0 |
| LICENSE | CC BY 4.0 |
| RIGHTS | CC BY |
| SOURCE | Europe PMC |