A new study published on arXiv investigates the reliability of explainable AI (XAI) in classifying freight transportation modes. Researchers used the PEPFAR Supply Chain Management System Delivery History dataset, comparing the performance of various machine learning models including neural networks, logistic regression, and random forests. The study found that tree ensembles generally outperformed other models in accuracy, and their feature importance rankings were more stable across retraining. However, significant disagreements were observed among models with similar accuracy regarding which features were most important, highlighting the need to assess explanation stability and faithfulness before relying on XAI for critical decisions. AI
IMPACT Highlights the need for rigorous evaluation of AI explanation methods before deployment in critical decision-making processes.
RANK_REASON Academic paper on explainable AI reliability. [lever_c_demoted from research: ic=1 ai=1.0]
- explainable AI
- Freight Mode Classification
- logistic regression model
- machine learning
- multilayer perceptron
- PEPFAR Supply Chain Management System Delivery History dataset
- random forest
- Shap
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →
