RedEx: beyond fixed representation methods via convex optimization (Q7017102)

Optimizing Neural networks is a difficult task which is still not well understood. On the other hand, fixed representation methods such as kernels and random features have provable optimization guarantees but inferior performance due to their inherent inability to learn the representations. In this paper, the authors aim at bridging this gap by presenting a novel architecture called RedEx (ReducedExpander Extractor) that is as expressive as neural networks and can also be trained in a layer-wise fashion via a convex program with semi-definite constraints and optimization guarantees. They also show that RedEx provably surpasses fixed representation methods, in the sense that it can efficiently learn a family of target functions which fixed representation methods cannot.\N\NFor the entire collection see [Zbl 1551.68029].