We combine ideas from machine learning, signal processing, optics, computer vision and physics to build better imaging systems (cameras, microscopes, and telescopes) through the co-design of optics, algorithms, and high-level tasks. Our aim is to design the next generation of smart, computational imagers that fuel scientific discovery, robotics, and medical diagnostics. We are particularly interested in:
- Differentiable optics - Can we use data and machine learning tools to design better cameras, microscopes, and telescopes?
- Physics-informed machine learning - How can we effectively combine our knowledge of imaging system physics with deep learning?
- Task-based imaging systems - What's the best camera or microscope for high-level tasks, such as robotics or medical diagnostics?
- Inverse problems and neural representations: Can we leverage neural priors to improve our imaging systems for microscopy and photography?
|Nov 2024||Kristina is giving an invited talk at MERL on Robust and Physics-informed machine learning for low-light imaging!|
|Nov 2024||Kristina is featured in an article on MIT News!|
|Jun 2023||Kristina will be joining Cornell’s Computer Sciences Department in Fall 2024 as an assistant professor!|