Dr. Patrick O'Driscoll is a Machine Learning Algorithm Development Engineer for Rebellion Photonics Houston, Texas. His research background is in machine learning methods for pattern recognition in large, complex, functional and temporal datasets. He currently research and develops real-time gas, fire, and intrusion detection methods for Rebellion Photonic's Gas Cloud Imaging safety system. He holds a B.S. in Chemical and Biomolecular Engineering, and Applied Mathematics and Statistics from Johns Hopkins University, USA, and an M.S. & Ph.D. in Applied Physics from Rice University, USA.
Recent advances in Machine Learning (ML), especially deep learning, have demonstrated superior to human performance for a variety of decision and recognition tasks. Together with advances in computational hardware and hyperspectral optics, affordable real-time ML based hazard event detection has become a reality. By providing the state-of-the art Artificial intelligence(AI) and ML based solutions for event detection, Rebellion Photonics embarks on a journey to revolutionize hazard and safety monitoring at all points of the petrochemical industry and beyond.
In this talk I will cover recent ML research efforts at Rebellion Photonics, with a focus on gas, fire and intrusion detection. Though comparative studies, advantages of data-driven algorithms are presented. Results from various field studies are also presented. It is demonstrated that ML based algorithms can adapt well to different environmental conditions, while improving its performance by learning over time.