The Jonas Lab uses machine learning to make scientific measurement faster, cheaper, and more powerful
The lab is lead by Eric Jonas and is part of the Department of Computer Science in the Physical Sciences Division at the University of Chicago . We are interested in using advances in machine learning and computational simulation to improve our ability to acquire and understand scientific data.
We are developing new techniques for structured prediction, Bayesian analysis and modeling, and deep learning for measurement.
Inverse problems underlie many aspects of measurement, ranging from computational imaging and microscopy to astronomy and spectroscopy.
We're interested in both deterministic and stochastic simulation at scale, scalable probabilistic inference, and serverless computing systems
Run linear algebra at scale using a serverless execution framework
Structured prediction for inverse problems
Run your code on thousands of cores with minimal overhead
Using machine learning to predict solar events
Or, ‘Could a neuroscientist understand a microprocessor?’
How can we exploit novel physics and computational algorithms to see through scattering media like fog and biological tissue?
How do we reverse-engineer the schematic of the brain? And how do we make sense of that data?
My Ph.D. thesis work, building probabilistic computing architectures to accelearte Bayesian computation.