Attribution Methods for Explainability of Predictive and Deep Generative Diffusion Models
| dc.contributor.advisor | Beakal Gizachew (PhD) | |
| dc.contributor.author | Debela Desalegen | |
| dc.date.accessioned | 2025-10-07T09:17:17Z | |
| dc.date.available | 2025-10-07T09:17:17Z | |
| dc.date.issued | 2025-06 | |
| dc.description.abstract | As machine learning models grow in complexity and their deployment in high-stakes domains becomes more common, the demand for transparent and faithful explainability methods has become increasingly urgent. However, most existing attribution techniques remain fragmented, targeting either predictive or generative models, and lack a hybrid approach that offers coherent interpretability across both domains. While predictive modeling faces challenges such as faithfulness, sparsity, stability, and reliability, generative diffusion models introduce additional complexity due to their temporal dynamics, tokento- region interactions, and diverse architectural designs. This work presents a hybrid attribution method designed to improve explainability for both predictive black-box models and generative diffusion models. We propose two novel methods: FIFA (Firefly-Inspired Feature Attribution), an optimization-based approach for sparse and faithful attribution in tabular models; and DiffuSAGE (Diffusion Shapley Attribution with Gradient Explanations), a temporally and spatially grounded method that attributes generated image content to individual prompt tokens using Aumann-Shapley values, Integrated Gradients, and cross-attention maps. FIFA applied to the Random Forest, XGBoost, CatBoost, and TabNet models in three benchmark datasets: Adult Income, Breast Cancer, and Diabetes, outperforming SHAP and LIME in key metrics: +6.24% sparsity, +9.15% Insertion AUC,-8.65% Deletion AUC, and +75% stability. DiffuSAGE evaluated on Stable Diffusion v1.5 trained on the LAION-5B dataset, yielding a 12.4% improvement in Insertion AUC and a 9.1% reduction in Deletion AUC compared to DF-RISE and DF-CAM. A qualitative user study further validated DiffuSAGE’s alignment with human perception. Overall, these contributions establish the first hybrid attribution methods for both predictive and\ generative models, addressing fundamental limitations in current XAI approaches and enabling more interpretable, robust, and human-aligned AI systems. | |
| dc.identifier.uri | https://etd.aau.edu.et/handle/123456789/7483 | |
| dc.language.iso | en_US | |
| dc.publisher | Addis Ababa University | |
| dc.subject | Diffusion models | |
| dc.subject | Explainable AI | |
| dc.subject | Feature attribution | |
| dc.subject | Integrated Gradients | |
| dc.subject | Shapley values. | |
| dc.title | Attribution Methods for Explainability of Predictive and Deep Generative Diffusion Models | |
| dc.type | Thesis |