AB  - The persistent threat of certificate fraud to the education sector underscores the urgent need for a robust and transparent fraud detection system. AI and machine learning techniques have proven effective in detecting fraudulent certificates but lack interpretability and transparency. High performing AI models often operate as ?black-boxes? making it difficult for users to trust or refine their decisions. This paper proposes a XAI-based fraud detection framework with human-understandable interpretation. The proposed framework was empirically validated on certificate behavioral log dataset. The implementation was carried out using Python environment. We employed XGBoost algorithm for classification, SMOTE to handle class imbalance and Shapley Additive Explanations (SHAP) to facilitate explainability. Experimental results demonstrate the efficacy of the proposed fraud detection framework and reduction in the number of false positives. SHAP technique quantify the contribution of behavioral features to model predictions. This will provide actionable insights for human review and institutional decision support. The performance of the proposed model shows that a robust ML-based mechanism for certificate fraud detection is practicable.
AU  - Khormi, I
AU  - Nanda, P
AU  - Mohanty, M
DA  - 2025/12/13
DO  - 10.1109/pcds68697.2025.11415273
EP  - 8
JO  - 2025 2nd International Symposium on Parallel Computing and Distributed Systems (PCDS)
PB  - Institute of Electrical and Electronics Engineers (IEEE)
PY  - 2025/12/13
SP  - 1
TI  - From Black Box to Transparent Gate: Explainable AI for Certificate Integrity
VL  - 00
Y1  - 2025/12/13
Y2  - 2026/06/07
ER  -