Speaker: Professor Eugenio Culurciello
Faculty Host: Professor Jonathan Chao
In this talk I will present our state-of-the-art work on neuromorphic hardware models of the mammalian visual system. In particular, I focus on modeling retinal pre-processing and the ventral visual pathway, with the goal of categorizing tens of objects from multi-megapixel cameras. We have designed and tailored multiple generation of bio-mimetic artificial retinas that can extract motion and contour information from the scene. We are developing light-invariant representations of the visual information for use in a hierarchical model of the ventral pathways, in particular V1, V2, V4 and IT. We have designed and implemented an innovative data-flow graphic-processing unit (GPU) - called VPU - Vision Processing Unit, that can instantiate very large neural networks (several million neurons), learn from data, and perform in real time and faster! We will show our implementation of convolutional neural networks in programmable digital hardware (FPGA) and on custom micro-chips (ASIC). Applications are in advanced modeling, neuroscience hypothesis testing, robotics, security, monitoring, cognitive hardware and sensor networks - to name a few.
About the Speaker
Eugenio Culurciello (S'97-M'99) received the Ph.D. degree in Electrical and Computer Engineering in 2004 from the Johns Hopkins University, Baltimore, MD. In July 2004 he joined the department of Electrical Engineering at Yale University, where he is currently an associate professor and directs the ‘e-Lab’ laboratory. His research interest is in analog and mixed-mode integrated circuits for biomedical instrumentation, synthetic vision, bio-inspired sensory systems and networks, biological sensors, silicon-on-insulator design. Eugenio Culurciello is the recipient of The Presidential Early Career Award for Scientists and Engineers (PECASE) and Young Investigator Program from ONR, the Distinguished Lecturer of the IEEE (CASS), and is the author of the book "Silicon-on-Sapphire Circuits and Systems, Sensor and Biosensor interfaces" published by McGraw Hill in 2009.