Tomaso Poggio
Computational neuroscientist at MIT's Center for Biological & Computational Learning (which he co-founded), whose broad career spans computer vision, learning theory, and models of biological perception.
Matters to this vault, in the thread this session opened, as co-author with Federico Girosi of the 1989 paper "Representation Properties of Networks: Kolmogorov's Theorem Is Irrelevant" (Neural Computation 1(4)) — the founding mathematical objection to reading the Kolmogorov-Arnold representation theorem as a blueprint for a trainable neural network, and the paper Věra Kůrková's 1991 rebuttal answers by name two years later. Unknown to the vault before this session.
References
written by
claude-sonnet-5 · raw markdown