analiza aiqbench
This research presents a structured workflow that integrates reference retrieval and contrastive SHAP analysis to improve the consistency and traceability of LLM-based risk diagnoses. By enforcing deterministic validation alongside evidence-grounded reasoning, the method addresses critical interpretability challenges in high-stakes machine learning applications.
A Controlled LLM Workflow for Evidence-Grounded Diagnosis of Risk Score Differences
arXiv:2610.23962v1 Announce Type: cross Abstract: We propose a controlled Large Language Model (LLM) workflow for explaining risk score differences using similar historical transactions. The workflow combines reference retrieval and contrastive SHapley Additive exPlanations (SHAP) evidence with reference-based LLM reasoning and deterministic validation. Experiments on a test dataset show that the workflow produces more consistent, evidence-grounded and traceable diagnoses than the same LLM without structured workflow guidance.
Abstract:We propose a controlled Large Language Model (LLM) workflow for explaining risk score differences using similar historical transactions. The workflow combines reference retrieval and contrastive SHapley Additive exPlanations (SHAP) evidence with reference-based LLM reasoning and deterministic validation. Experiments on a test dataset show that the workflow produces more consistent, evidence-grounded and traceable diagnoses than the same LLM without structured workflow guidance.
Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
arXiv:2610.23962 [cs.AI]
(or arXiv:2610.23962v1 [cs.AI] for this version)
https://doi.org/10.48550/arXiv.2610.23962 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
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 · 158 cuvinte