AI for Science and Applications

The AI for Science and Applications research area investigates how AI can accelerate scientific discovery and enable data-driven solutions to complex societal challenges. Research in this area develops machine learning methods that contribute across the scientific workflow, including hypothesis generation, predictive modeling, experimental design, and decision-making in scientific and engineering domains. A central focus is on building AI systems that can learn from complex, heterogeneous, and often sparse scientific data; integrate physical models and domain knowledge; and produce reliable, interpretable insights under uncertainty.

Faculty in this area work across the full pipeline of scientific inquiry, including extracting structure and anomalies from high-dimensional data, predicting properties of molecules and materials, designing new compounds and systems, and guiding adaptive experiments in automated and robotic laboratories. These efforts are coupled with advances in representation learning, statistical signal processing, graph and structured-data modeling, and trustworthy AI, enabling models that are robust, data-efficient, and aligned with scientific reasoning.

Applications span a wide range of domains, including microbiome and multi-omics science, molecular and materials discovery, environmental and climate systems, infrastructure and power-grid forecasting, and other areas where integrating data-driven methods with physical and domain-based models is essential. By tightly connecting methodological innovation with domain collaboration, this research area aims to transform how scientific knowledge is generated, validated, and translated into real-world impact.