BEGIN:VCALENDAR VERSION:2.0 PRODID:-//Memento EPFL// BEGIN:VEVENT SUMMARY:AI-enhanced vision: seeing the invisible DTSTART:20200316T121500 DTEND:20200316T130000 DTSTAMP:20260509T215419Z UID:2c4e1956ccaea95591b4c2ada80469cc4028ab97a1ed365c2cccac6b CATEGORIES:Conferences - Seminars DESCRIPTION:Prof. Dr. George Barbastathis\,\nMIT\nInstitute of Microengine ering - Distinguished Lecture\n\nUpdate as of 13 March 2020: Due to travel restrictions\, we regret that the speaker will not be able to travel to E PFL. However\, the lecture will be held remotely by zoom. No transmission will be organized to lecture halls\, but participants can join remotely vi a the zoom link provided below. \n\nCampus Lausanne: the talk will not be available in the lecture hall on EPFL Campus.\nCampus Microcity: the talk will not be transmitted to a lecture hall in Microcity.\n\nPlease use the zoom link below to join remotely:\nZoom Live Stream: https://epfl.zoom.us/ j/506874457\n\nAbstract: If you point your camera to a scene\, and the cam era registers nothing—does it mean that nothing was really there? Hardly ! The camera pixels measure “raw” light intensity where the encoded in formation often is much richer than a human observer could tell just by lo oking at the pixels on a screen. Which algorithms\, then\, should one appl y to decode the raw intensity and reveal the hidden scene?\nIn this semina r\, I will describe how to use Deep Neural Networks (DNNs)\, a form of Mac hine Learning (ML) algorithm\, to perform this decoding. During the traini ng stage of the DNN\, physically generated objects are used to produce the encoded raw intensities. From these pairs of objects and raw intensities the DNN learns the association between the scenes and their encoded repres entations. After training\, given a new scene\, the DNN decodes it correct ly to produce a final reconstructed image that is meaningful to a human ob server.\nWith my research group\, we applied this approach to three challe nging instances of invisibility: transparent objects\, also known as “ph ase objects\,” whose raw intensities are highly rippled diffraction patt erns\; phase objects that are also very dark\, i.e. the diffraction patter ns are also highly attenuated\; and objects hidden behind or surrounded by diffusers\, e.g. frosted glass or multiple layers of glass patterned with sharp light-scattering features.\nIt is important to emphasize that in ou r work ML is not used in the traditional way to interpret the scenes\; rat her\, it is used to form interpretable representations of scenes in situat ions where traditional ML would be helpless due to physical limitations in the optics. The cooperation of ML with physical models proved to be very powerful in this work and\, beyond\, is certain to impact many fundamental and applied aspects of physical and life sciences and engineering.\n\nBio : George Barbastathis received the Diploma in Electrical and Computer Engi neering in 1993 from the National Technical University of Athens (Πολυ τεχνείο) and the MSc and PhD degrees in Electrical Engineering in 1 994 and 1997\, respectively\, from the California Institute of Technology (Caltech.) After post-doctoral work at the University of Illinois at Urban a-Champaign\, he joined the faculty at MIT in 1999\, where he is now Profe ssor of Mechanical Engineering. He has worked or held visiting appointment s at Harvard University\, the Singapore-MIT Alliance for Research and Tech nology (SMART) Centre\, the National University of Singapore\, and the Uni versity of Michigan – Shanghai Jiao Tong University Joint Institute (密 西根交大學院) in Shanghai\, People’s Republic of China. His resear ch interests are in machine learning and optimization for computational im aging and inverse problems\; and optical system design\, including artific ial optical materials and interfaces. He is member of the Society for Phot o Instrumentation Engineering (SPIE)\, the Institute of Electrical and Ele ctronics Engineering (IEEE)\, and the American Society of Mechanical Engin eers (ASME). In 2010 he was elected Fellow of the Optical Society of Ameri ca (OSA) and in 2015 he was a recipient of China’s Top Foreign Scholar ( “One Thousand Scholar”) Award.\n\nNote: The Seminar Series is eligible for ECTS credits in the EDMI doctoral program\n\nNote: After the lecture\ , there will be time for discussion and interaction with the distinguished speaker\, sandwich lunch and refreshments sponsored by the Institute of M icroengineering will be provided for attendees in front of the lecture hal l (BM 5104\, ca. 13h15) LOCATION:ONLINE ONLY https://epfl.zoom.us/j/506874457 STATUS:CONFIRMED END:VEVENT END:VCALENDAR