Research

Graph methods for language, signals, biomedical data, and public-sector analytics.

1. Graphs and spectral methods

• Discrete magnetic Laplacians.
• Spectral clustering.
• Isospectral graphs and spectral bracketing.
• Isospectral graphs via spectral bracketing
• Spectral preorder and perturbations

2. Graph signal processing

• Entropy and complexity metrics for graph signals.
• Applications to EEG, fMRI, sensors, and two-phase flow data.
• Permutation entropy for graph signals
• Dispersion entropy for graph signals
• Graph-based permutation patterns for fMRI

3. NLP and graph representation learning

• NLP models using graph representation learning and reasoning.
• GPU-accelerated training and evaluation workflows in PyTorch.

4. Public-sector analytics

• Nationwide assessment data.
• Multilevel modelling and graph-based clustering.
• GIS equity maps and Shiny dashboards for non-technical policymakers.
• Decision-support tools that reduced analysis-to-decision time from weeks to hours.