Independent Research Groups / Neural Data Structure and Representation
We develop machine learning tools for analyzing, representing, and visualizing high-dimensional neuroscience data such as single-cell omics, clinical images, or time series.
Neural Data Structure and Representation
Our group studies how machine learning can reveal structure in high-dimensional neuroscience and biomedical data. Much of this data is unlabelled, either because it arises in exploratory research settings or because annotation is prohibitively costly. We therefore develop and analyze unsupervised and self-supervised approaches that generate reliable insights without labels. Our work focuses on informative representations, clustering, dimensionality reduction, and visualization, guided by principled ideas from geometric deep learning and topological data analysis. We apply these methods across diverse modalities, including single-cell omics, clinical imaging, and neuronal time series to better understand brain health data.
