Our group combines rigorous modeling, optimization, and machine learning to create methods with both theoretical depth and practical impact.
Build elegant theory and powerful algorithms that merge physics, optimization, and AI to solve high-impact imaging and sensing problems.
Physics-aware inverse methods, compressive sensing, coded sensing, and learning-based reconstruction for next-generation imaging systems.
Explore direction →Satellite and airborne LiDAR, waveform and photon-counting systems, terrain and canopy reconstruction, and computational Earth observation.
Explore direction →Diffusion models, VAEs, GANs, uncertainty-aware inference, quantile learning, and accelerated reconstruction for scientific imaging.
Explore direction →Tensor-based hypergraph learning, directed hypergraphs, graph neural networks, and structured representations for complex relational data.
Explore direction →Research with scientific reach and real-world relevance.
Developing computational sensing and generative reconstruction methods for photon-limited, height-resolved Earth observation, with applications to vegetation structure, topography, and geospatial mapping.
Combining sensing, compression, and machine learning to improve quality, speed, and bandwidth efficiency in imaging pipelines.
Building mathematical and algorithmic foundations for tensor-based hypergraph representations and learning systems.
Creating robust learning frameworks that fuse domain physics, optimization, and modern AI for hard inverse problems.
Recent papers (2023–2026) organized by the group's four research directions, spanning computational imaging, remote sensing and LiDAR, generative AI for inverse problems, and hypergraph signal processing.
The broader record includes books, patents, invited talks, editorial leadership, and a large body of journal and conference publications across signal processing, optics, imaging, remote sensing, and machine learning.