Advancing AI Through Statistical Foundations
The Statistics and Data Science Department at MBZUAI advances the frontiers of artificial intelligence through rigorous statistical methodology, scalable computation, and interdisciplinary collaboration. Our research bridges theoretical statistics, machine learning, and domain applications to build AI systems that are principled, interpretable, and impactful. We develop methods that combine statistical rigor with computational scalability to address challenges in climate science, health, engineering, and other data-intensive domains. Every project reflects MBZUAI’s commitment to advancing AI responsibly—grounded in scientific integrity, ethical standards, and societal benefit.
Research Themes
Probabilistic Modeling and Forecasting
We design models that quantify and communicate uncertainty, enabling informed decision-making under changing and uncertain conditions. Our research spans probabilistic machine learning, time-series modeling, anomaly detection, and forecast evaluation—capturing nonlinear dynamics, seasonality, and heterogeneity in complex systems.
Optimization and Scalable Learning Systems
Our faculty develop efficient and robust optimization algorithms that empower modern machine-learning systems to operate at scale. Research areas include stochastic and distributed optimization, privacy-preserving learning, and large-scale model training under real-world constraints such as decentralization and noise.
Trustworthy and Reliable AI
We are committed to making AI systems more interpretable, fair, and dependable. Through statistical calibration, uncertainty quantification, causal reasoning, and model validation, our work ensures that AI models are transparent and reliable across diverse environments and applications.
Graph and Network Intelligence
Our researchers explore the interconnected nature of data using graph-based learning and representation techniques. By modeling relationships among entities, we reveal insights about complex social, technological, and scientific networks, advancing fundamental understanding of connected systems.
Applied AI for Science and Industry
Grounded in rigorous methodology, we translate AI and statistical innovations into real-world impact. Applications range from climate and environmental forecasting to algorithmic trading, user-growth analytics, and materials discovery—demonstrating how data-driven intelligence can drive progress across disciplines.
Collaborative Ecosystem
Our department fosters collaboration across academia, government, and industry. Working closely with MBZUAI’s AI centers and global partners, we translate research into actionable solutions that strengthen the reliability, efficiency, and sustainability of AI in society.
Featured Projects
Climate Model Emulation with Neural Operators
ActiveFast and accurate emulation of complex climate models using physics-informed neural networks
Causal Discovery in High-Dimensional Systems
ActiveNovel algorithms for learning causal structures from observational and interventional data
Uncertainty Quantification for AI Systems
ActiveMethods for reliable uncertainty estimates in deep learning models for critical applications
Extreme Event Prediction with Spatio-Temporal Models
NewStatistical and ML approaches for forecasting rare weather events and natural disasters
Selected Publications
Neural Operator Approaches for Climate Model Emulation
Nature Climate Change
Causal Structure Learning in Non-stationary Environments
ICML 2025
Uncertainty Quantification in Deep Bayesian Networks
NeurIPS 2024
Spatio-Temporal Modeling of Extreme Weather Events
JASA
Theoretical Analysis of Transformer Architectures
JMLR
Open Collaborations
Interested in Collaborating?
We welcome partnerships with academic institutions, research labs, and industry organizations
Our department actively seeks collaborative opportunities in:
- Joint research projects and grants
- Student and faculty exchange programs
- Industry partnerships for applied research
- International workshops and conferences