C3: AI-Driven Active Learning for Predictive Materials Design

AI-Driven Active Learning for Predictive Materials Design

This doctoral project aims to develop artificial intelligence methods that accelerate the discovery and optimization of functional thin films through active learning. Building on the Sparse Symbolic Inference (SSI) framework, the candidate will train interpretable models that uncover simple, physically meaningful relationships between synthesis parameters, microstructure, and functional performance. Using features derived from microscopy and spectroscopy data, these models will predict key material properties and identify promising, unexplored regions in the materials space. Novel feature extraction methods—such as binary encoding for compressed representations—will be employed to handle sparse and heterogeneous datasets. The resulting predictive rules will form the basis of a closed-loop optimization workflow that guides targeted synthesis and characterization. This approach will enable data-efficient, explainable AI for materials design, linking microscopic descriptors with macroscopic functionality.

Supervisor

s