Rafael Orozco

Rafael Orozco

Researcher in imaging with uncertainty and machine learning

About me

Email: rorozco@gatech.edu

I received by Phd from the Seismic Laboratory Imaging and Modeling and am working at the Subsurface Innovation Lab in Oxy.

My research is in machine learning for acceleration and uncertainty quantification of image reconstruction. During my PhD research, I implemented generative models that learn to sample from the Bayesian posterior of realistic high-dimensional imaging problems. My engineering mindset drives me to bridge the gap between innovative algorithms and practical applications, making these solutions impactful for industry needs.

My main applications are:

seismic imaging and digital twin data assimilation for monitoring of carbon dioxide (CO2) sequestration to mitigate climate change…

AN UNCERTAINTY-AWARE DIGITAL SHADOW FOR UNDERGROUND MULTIMODAL CO2 STORAGE MONITORING

“BEACON: Bayesian Experimental design Acceleration with Conditional Normalizing flows - a case study in optimal monitor well placement for CO2 sequestration”

“Inference of CO2 flow patterns – a feasibility study”

…and various medical imaging modalities including X-ray CT

Medical Imaging with Uncertainty Quantification

“Adjoint operators enable fast and amortized machine learning based Bayesian uncertainty quantification”

Photoacoustic imaging

Medical Imaging with Uncertainty Quantification

“Photoacoustic imaging with conditional priors from normalizing flows”

Full waveform inversion with ultrasound for intracranial imaging

“ASPIRE: Iterative Amortized Posterior Inference for Bayesian Inverse Problems”

and optimal experimental design for accelerated Magnetic Resonance Imaging

“Probabilistic Bayesian optimal experimental design using conditional normalizing flows”

Generally, I focus on imaging governed by computationally expensive partial differential equations (PDEs). These modalities are best suited for physics hybrid frameworks that are accelerated by machine learning but informed by traditional domain physics knowledge in the form of PDE simulations. I primarily consider myself an engineer since my passion is in scaling these novel techniques to real world applications.

Recent News