Dr. AlRegib's research on machine learning and image/video processing has important implications for fields like autonomous vehicles, medical image analysis, computational seismic interpretation, and computational ophthalmology.

His Omni Lab for Intelligent Visual Engineering and Science (OLIVES) group approaches their research from two avenues: robust learning and learning with limited labels. Dr. AlRegib and his team develop algorithms that can robustly operate under challenging, real-world conditions through weakly supervised learning, back-propagated gradients, hyperpolar classification, and transfer learning. They compiled several large-scale datasets to test and develop these algorithms, which will advance industry areas like autonomous driving, remote repositioning, smart and connected health care, defense models design, and computational seismic interpretation. 

The OLIVES group also learns to characterize data using limited labels via weakly or semi-supervised learning and sequence modeling. In this area, Dr. AlRegib and his researchers have developed important datasets as well as an interactive tool for salt interpretation benchmarking in large subsurface volumes. The group has been known in the community as the first and the lead in developing effective machine learning methods for seismic interpretation. Limited label techniques hold significant potential for applications such as subsurface lithology, structure and stratigraphy characterization, material characterization, optical coherence tomography (OCT) analysis, fundus images analysis, and medical imaging.  

Research Goals  

  • Autonomous vehicles: Improving scene and object recognition algorithms for more effective autonomous navigation, even in challenging environments 
  • Smart transporation: Improving infrastructure intelligence
  • Seismic imaging: Advancing practices for processing geosignals and energy data by overcoming current obstacles to machine learning integration like conflicting research labels 
  • Healthcare intelligence: Enhancing healthcare via intelligent tele-healthcare and machine learning for more effective human decisions  

Activities 

  • Material characterization: Developing computational models for characterizing apparent or latent material properties to support machine object recognition and scene understanding 
  • Scene recognition: Creating and testing algorithms for object and scene recognition under challenging conditions and compiling large datasets to analyze robustness 
  • Seismic interpretation: Leveraging deep learning techniques to characterize data using limited labels, improving techniques for tasks like facies classification 
  • Intelligent eye care: Developing effective machine learning modules and systems that can revolutionize the eye healthcare sector with the ultimate goal of preventing blindness

Leadership  

  • Director, Omni Lab for Intelligent Visual Engineering and Science (OLIVES) , Georgia Tech 
  • Director, Center for Signal and Information Processing (CSIP), Georgia Tech 
  • Director, Center for Energy and Geo Processing (CeGP), Georgia Tech 
  • Director, Initiatives and Programs in the Middle East and North Africa (MENA), Georgia Tech 
  • Technical Program Chair, IEEE International Conferences
  • IEEE Senior Member