Research

Optical sensors often produce complex signals that contain information about several physical parameters at once. Combining photonic design with machine learning offers a way to interpret these responses, distinguish overlapping effects, and extract useful measurements. This research direction explores how data-driven methods can expand the capabilities of optical sensing across different devices and applications.

Our focus includes multi-parameter estimation, feature extraction, and the integration of physical insight into learning models. We examine how sensor design and signal analysis can complement one another, with particular attention to limited training data and reliable performance under varying conditions. Through this approach, we aim to turn rich optical responses into practical tools for intelligent sensing.