explainable-ai

This paper proposes BREAD, a scalable baseline-referenced explanation method for diagnosing anomalies in AI-based statistical process monitoring, with theoretical guarantees showing higher faithfulness than LIME under mean-shift anomalies.

This empirical study compares conversational XAI (powered by LLMs) against a traditional dashboard for UAV intrusion detection auditing, finding the conversational interface improves perceived usefulness but risks operator over-reliance on AI advice.

Introduces SPOT (Sampling Policy Observation Tree), a model-agnostic framework for interpreting deep reinforcement learning policies by constructing finite-horizon lookahead trees that expose downstream consequences of actions, demonstrated on SUMO-RL traffic-signal control.

This position paper advocates computational argumentation as a formal foundation for Evaluative AI, which supports human decision-making by presenting competing hypotheses with evidence for and against, rather than single recommendations.

Introduces LiFTER, a neuro-symbolic predictor for continuous-time dynamic graph forecasting that grounds predictions in observable temporal facts and executable rules, enabling fully inspectable and verifiable link prediction with competitive accuracy and high explanatory fidelity.

The paper investigates whether guideline-based categorical encodings of continuous predictors can replace continuous inputs in stroke outcome prediction models without sacrificing accuracy, finding comparable performance in most treatment cohorts.

This paper proposes Patients-like-me (PLM), a unified LM–GNN framework that integrates local patient semantics with global cohort structure for explainable clinical prediction. It introduces a Variational Expectation-Maximization algorithm and demonstrates state-of-the-art results on MIMIC-III and MIMIC-IV with reference-patient explanations.

The paper introduces a distribution-based framework to measure the stability of attribution methods (explainers) by quantifying the separability of feature rankings and identifying the maximum top-k ranking that remains reliable across stochastic runs.

A systematic review of XAI research in the context of the EU Right to Explanation, analyzing gaps between legal requirements and technical implementations across GDPR and the AI Act.

A new MIT-led study in Nature Medicine finds that AI assistance and explainability methods impact skin disease diagnosis accuracy differently depending on user expertise: non-experts over-trust AI explanations, while clinicians perform best with only the model's prediction. The results highlight the need for user-centered AI design that accounts for automation bias.

This paper introduces xMICD, a method that combines ICD code groupings with pre-trained embedding similarities to create low-dimensional, clinically interpretable patient representations, achieving predictive performance comparable to embedding-based approaches.

This paper applies explainable machine learning to mobility trajectories from the NetMob 2025 Data Challenge to assess the 15-minute city concept in Paris, finding that higher local service availability is associated with less car use and more active mobility, though with spatial and demographic heterogeneity.

This paper introduces a white-box, gradient-regularized evasion framework that embeds attack logic directly into model parameters, successfully fooling explainable AI auditors like LIME, SHAP, and Integrated Gradients while bypassing anomaly detection defenses.

This paper proposes an automated multilabel classification system for Mpox research articles using BERT, achieving 97% accuracy, and employs SHAP for explainability. The system aims to help researchers and healthcare workers quickly find relevant information.

HyCE-RAG is a novel hypergraph-based retrieval-augmented generation framework for multi-hop question answering that constructs explicit evidence chains via confidence-aware heuristic search, outperforming standard RAG and graph-based RAG methods in accuracy, relevance, and faithfulness.

This paper systematically evaluates how post-training quantization affects the interpretability of CNN models using Grad-CAM and LIME, revealing that classification accuracy is not a reliable indicator of interpretability stability and that architecture selection is critical for trustworthy deployment.

Introduces CEL, a unified library and benchmark for counterfactual explanations, providing standardized implementations of 14 methods across 18 datasets to enable fair comparison in explainable AI.

This paper presents an Automated Data Processing (ADP) framework that uses reinforcement learning and Shapley-based explainable AI to optimize machine learning model-feature combinations for warpage detection in fused deposition modeling, achieving improved accuracy and stability.

Proposes a domain-agnostic framework for generating grounded natural language explanations for time series forecasts using large language models, reducing hallucination by constraining to verifiable evidence. Evaluated on financial and freight pricing case studies.

This paper presents FST.ai 2.5, an explainable and uncertainty-aware AI framework for Olympic and Para-Taekwondo that integrates athlete digital twins, competition analytics, and federation-scale decision support.