Event:

Object recognition is a difficult computational problem. Nevertheless, the human visual system can rapidly and effortlessly recognize objects in cluttered scenes under widely varying viewing conditions at a level beyond that of current machine vision systems. In the past several years, we have made significant progress in understanding the neurocomputational mechanisms underlying rapid object recognition in cortex. Remarkably, recent results covering areas as varied as face perception, reading, car categorization and auditory phoneme processing suggest an appealingly simple unified account of how our brains assign meaning to sensory stimuli, based on a simple-to-complex? processing hierarchy that builds invariant feature representations which in turn form efficient representations for downstream modules tuned to different recognition tasks. Specifically, our HMAX model of object recognition in cortex has proved to not just be a tool to summarize insights from neuroscience regarding how the brain recognizes objects and to drive new experiments, but it has also turned out to be a versatile framework to develop neuromimetic machine vision systems. A particular advantage of the neuromimetic approach to machine vision is that its firm grounding in neuroscience research promises further advances in capabilities as we continue to improve our understanding of how the brain sees. In my talk, I will review the HMAX model and the neuroscience research it is based on, present benchmark results, and will describe current research designed to further improve its capabilities.

Bio: Dr. Riesenhubers main research foci are the neural mechanisms underlying object recognition and task learning in the human brain, and their translation to machine vision systems as well as hybrid brain-machine object detection systems. The computational model at the core of his research has been quite successful in elucidating the neural mechanisms underlying robust invariant object recognition, contributing to Technology Review Magazines naming him one of their TR100 in 2003, the 100 people under age 35 whose contributions to emerging technologies will profoundly influence our world. In addition to its success in neuroscience, Dr. Riesenhubers model of biological vision is also currently achieving state-of-the-art results on several benchmark object detection tasks in machine vision, leading a number of machine vision groups in the US and abroad to adopt this model for their work. Dr. Riesenhuber has received several awards, including a McDonnell-Pew Award in Cognitive Neuroscience and an NSF CAREER Award. He holds a PhD in computational neuroscience from MIT.


Room: Raum 03.07.023 (SEMINAR (INF 6/13))
see: https://portal.mytum.de/displayRoomMap?roomid=03.07.023@5607&disable_decoration=yes