Tamas Budavari
Johns Hopkins University, MD·Astronomy & Astrophysics
Telescopes that image the same patches of sky over and over now record huge numbers of objects that change in brightness, from regularly pulsing stars to sudden explosions. Sorting through this flood of data to find, measure and classify these variable and transient objects is a central challenge.
- Fields
Astronomy & Astrophysics
Mathematics & Statistics
Computer Science
- Topics
astronomy
astrostatistics
sky surveys
- Grants
- 1 known grant, through Aug 2027
What they work on
- Develop algorithms that classify variable and transient astronomical objects from the shapes of their multiband light curves, records of brightness over time in several colors, using multi-epoch survey data.
- Build Bayesian neural network classifiers, which report their own uncertainty, with improved priors and better exploration of possible solutions.
- Improve how source likelihood functions and source uncertainties are represented to enable optimal detection of faint variable objects and demographic modeling of cosmic populations.
More about the research
It matters because only a small fraction can be followed up with other telescopes. This collaborative project, joining astronomers, applied mathematicians and statisticians, will develop statistical and machine learning methods and open-source software for multi-epoch surveys, meaning surveys that observe the sky repeatedly.
Why it matters
Sky surveys produce more variable and transient objects than astronomers can study in detail. Better detection and classification tools could help them spend limited telescope follow-up time on the most important objects. Because all algorithms will be released as open-source software, other surveys could use them to extract more science from their data.
Known grants
From NIH and NSF research grants. Labs often have other funding we can't see.
Closest labs
Vanessa Chen
Carnegie Mellon University, PA·Engineering·Early-career
Design bio-inspired, time-domain analog classifiers that exploit transistor mismatch and nonlinearity to perform energy-efficient hierarchical classification at low supply voltages within a sensor.
Andrew Mann
University of North Carolina at Chapel Hill, NC·Astronomy & Astrophysics·Early-career
Measure luminosity, radius, mass and age to better than 5% accuracy for 5–500 million-year-old K and M stars by analyzing light curves of eclipsing and astrometric binaries found by Gaia, K2 and TESS.
Susanta Ghosh
Michigan Technological University, MI·Engineering·Early-career
Develop a Bayesian convolutional neural network that predicts electron density, the map of where electrons reside in a material, while reporting its own uncertainty.
The school
Johns Hopkins University: 100 labs on PI Pages, $4.1B a year on research, 622 research doctorates a year.
Reach out
A start for your email. Fill in the brackets.