Abstract

In the era of pervasive data-driven decision-making, statisticians increasingly integrate machine learning (ML) and artificial intelligence (AI) techniques to tackle complex, high-dimensional, and dynamic datasets. While statistics traditionally focuses on understanding past phenomena through probabilistic inference and human-guided analysis, ML excels at prediction and pattern discovery via algorithmic optimization, and modern AI—especially neural architectures and agentic systems—pursues generalizable, autonomous intelligence. These differences offer rich opportunities for synergy: statistical principles can improve ML/AI rigor, transparency, and uncertainty handling, while ML/AI innovations expand statistical capabilities in scalability, automation, and real-world deployment. This session invites contributions that advance this convergence, providing fresh insights to strengthen all three disciplines. It aims to cultivate transparent, predictable, and ethically sound pathways from data to actionable conclusions. Relevant topics include (but are not limited to):

1. Innovative statistical foundations for ML/AI algorithms, including robust inference, uncertainty quantification, and causal methods in predictive modeling.
2. Interdisciplinary applications of integrated statistical-ML-AI approaches in domains such as healthcare, finance, environmental modeling, social sciences, and official statistics.
3. Ethical frameworks, fairness, bias mitigation, transparency, and accountability when embedding statistical methods in ML/AI systems.
4. Statistical advances in data quality, preprocessing, feature engineering, and handling missing or non-i.i.d. data to boost ML/AI performance.
5. Enhancing explainability and interpretability of ML/AI models through statistical tools, including post-hoc methods, counterfactuals, and inherently interpretable hybrids.
6. Recent developments in statistical learning, ensemble methods, deep learning integration, reinforcement learning, and emerging paradigms like agentic AI.
7. Hybrid modeling strategies that combine classical statistical techniques with ML/AI to exploit complementary strengths for superior accuracy, efficiency, and trustworthiness.

The session welcomes researchers, practitioners, and educators to share theoretical advances, methodological innovations, case studies, and empirical lessons.