Principal Research Scientist
CSAIL (Computer Science and Artificial Intelligence Lab)
MIT Computer Vision & Graphics Group
Investigator, Athinoula A. Martinos Imaging Center
MIT, 77 Massachusetts Avenue
Cambridge, MA 02139
Phone: 617 452 2492
After a French baccalaureate in Physics and Mathematics and a B.Sc. in Psychology (minor in Philosophy), Aude Oliva received two M.Sc. degrees –in Experimental Psychology, and in Cognitive Science and a Ph.D from the Institut National Polytechnique of Grenoble, France. She joined the MIT faculty in the Department of Brain and Cognitive Sciences in 2004 and the MIT Computer Science and Artificial Intelligence Laboratory - CSAIL - in 2012.
Her research is cross-disciplinary, spanning human perception/cognition, computer vision, and cognitive neuroscience, focusing on research questions at the intersection of the three domains. Her work in Computational Perception and Cognition builds on the synergy between human and machine perception and cognition, and how it applies to solving high-level recognition problems like understanding scenes and events, perceiving space, localizing sounds, recognizing objects, modelling attention, eye movements and visual memory, as well as predicting subjective properties of images (like image memorability). Her research integrates knowledge and tools from image processing, image statistics, computer vision, human perception, cognition and neuro-imaging (fMRI, MEG). Her work has been regularly featured in the scientific and popular press, in museums of Art and Science as well as in textbooks of Perception, Cognition, Computer Vision and Design. She is the recipient of a National Science Foundation CAREER Award. Her research programs are funded by the National Science Foundation, the National Eye Institute, Google and Xerox. Her Curriculum-Vitae (pdf), her google scholar profile page.
Cross-disciplinary research bridges the gaps from theory to experiments to applications, accelerating the rate at which discoveries are made by solving problems through a novel way of thinking. To this end, my research in human perception and cognition span three disciplines:
- Psychology: We employ psychophysics and behavioral methods to discover phenomena of human perception and cognition.
- Computer Vision: We use methods in image processing and computer vision as tools to model and predict psychological phenomena as well as to provide computer vision with new applications.
- Human Cognitive Neuroscience: Armed with theoretical and computational frameworks, we use brain imaging experiments (e.g., fMRI, MEG) to study how the human brain represents perceptual and cognitive phenomena.
My work has capitalized on visual scene understanding. Scene understanding is a multidisciplinary field, but the traditional layered structure of academia and publication outlets have created substantial barriers between researchers working on human perception and computer vision. In my work I have constantly sought to bridge these two areas of research. Building on foundations in image statistics and image processing, I have developed new experimental paradigms to study human perception and memory, and novel methods in human neuroscience. I have integrated ideas from cognitive psychology into computer vision and I have applied computer vision to model various ecological tasks like scene classification, depth perception, visual search and memory. I have also established successful bridges with computer vision, contributing to its awareness of studies in psychology. Some of my earlier contributions (i.e., hybrid images) have also made a strong impact outside the field of human perception. My 2001 paper in Int. J. of Computer Vision, which introduced the first holistic computational model of scene recognition, and what would later be referred as GIST features, has generated much follow-up work in both human vision and computer vision. Recently, I introduced a new domain of application of computer vision: memorability. The study of long-term memory provides another avenue of research in which to probe what kind of information is extracted, stored, and used to predict behaviors.
In order to satisfactorily study the dynamics of perception, cognition and action in the human brain, measurements at the level of milliseconds and millimeters are required. Combining the best of two worlds (ms-resolution MEG and mm-resolution fMRI) allows to study, in a non invasive manner, how recognition processes unfold concurrently in time and space in the human brain: In a first study in Nature Neuroscience 2014 (MIT news release), we have applied the new approach to visual object recognition.
List of Publications here
As an undergraduate student, I studied the Philosophy of Existentialism: it teaches you "fall seven times, stand up eight" (Japanese proverb). Building cross-disciplinary knowledge in psychology, computer vision and neuroscience took me twenty years (and I will never stop learning). I look forward to the next few decades to the chance to teach students how to become trans-disciplinary thinkers. I have witnessed too many times that, “at every crossway on the road that leads to the future, each progressive spirit is opposed by a thousand men appointed to guard the past,” (A.G. Bose) but also that, “logic will get you from A to Z; imagination will get you everywhere.” (A. Einstein). I believe that cross-disciplinary education opens people’s minds to imagine, believe in and do, the impossible.