Professor
Tomaso Poggio
Massachusetts Institute of Technology
Neuroscientist; Educator
Area
Biological Sciences
Specialty
Neurosciences
Elected
1997
The problem of (human) intelligence is, I believe, the greatest problem in science. My scientific goal is to advance the science, first, and the engineering, second, of intelligence.This is also the ambitious aim of the Center for Brains, Minds, and Machines of which I am the Director.
By integrating technological breakthroughs in neuroscience, continued rapid advances in computer power, and accumulated knowledge in machine learning and cognitive science, we hope to make a significant leap in our scientific understanding of human intelligence and its neural basis. In my view, this focus on integration, reflected in our name—the Center for Brains, Minds, and Machine—is essential: I aim to bring together in the Center computer scientists, cognitive scientists and neuroscientists.
My personal research focus is on theory of learning and on human vision. In machine learning the mathematical questions I am working on are about the power of deep learning architectures. Under which conditions are deep networks better than shallow? We have established interesting connections between learning of real and Boolean functions, and between sparsity and learning of hierarchical, compositional functions.
On human vision we are building on i-theory, that we developed over the last 4 years, to simulated hierarchical architectures that reproduce human vision abilities and deficiencies unlike existing deep learning networks. The theory seems able to connect invariance properties of sensory information to the architecture of sensory cortex and to the properties of single neurons in the ventral stream of visual cortex.
By integrating technological breakthroughs in neuroscience, continued rapid advances in computer power, and accumulated knowledge in machine learning and cognitive science, we hope to make a significant leap in our scientific understanding of human intelligence and its neural basis. In my view, this focus on integration, reflected in our name—the Center for Brains, Minds, and Machine—is essential: I aim to bring together in the Center computer scientists, cognitive scientists and neuroscientists.
My personal research focus is on theory of learning and on human vision. In machine learning the mathematical questions I am working on are about the power of deep learning architectures. Under which conditions are deep networks better than shallow? We have established interesting connections between learning of real and Boolean functions, and between sparsity and learning of hierarchical, compositional functions.
On human vision we are building on i-theory, that we developed over the last 4 years, to simulated hierarchical architectures that reproduce human vision abilities and deficiencies unlike existing deep learning networks. The theory seems able to connect invariance properties of sensory information to the architecture of sensory cortex and to the properties of single neurons in the ventral stream of visual cortex.
Last Updated