Shubham Bhardwaj

Introduces PHAST, a port-Hamiltonian-based architecture for stable long-horizon forecasting from position-only observations, decomposing Hamiltonian components and demonstrating results across thirteen q-only benchmarks.

Multimodal clustering study revealing circuit-level signatures in Parkinson's disease (bioRxiv preprint, co-author).

Paper on a curvature-informed strategy for furthest point sampling in 3D data.

Conference paper (ICONIP 2024) and arXiv posting on using hyperspectral data and sparse spatio-temporal learning for robust night-time aerial material segmentation.

Patent application for multi-attribute synthetic data generation for fashion/apparel and related annotation/synthesis systems, assigned to Jio Platforms Ltd; lists Shubham Bhardwaj as an inventor.

Patent application describing an automated imaging system for extracting object footprints and other geospatial assets from imagery, assigned to Jio Platforms Ltd; lists Shubham Bhardwaj as an inventor.

Master's thesis work at UT Austin (2022) involving Gaussian processes and related machine learning topics, supervised by Chandrajit Bajaj.