Amit Daniely

activity

20152023

most citedNeural Networks Learning and Memorization with (almost) no Over-Parameterization

17 citations · 72 across the 15 of their papers we have counts for

collaborators

Showing 2017 · cs.LGShow all

3 papers · 2 filters

cs.LG2017★ 7 cited

Random Features for Compositional Kernels

Amit Daniely, Roy Frostig, Vineet Gupta +1

We describe and analyze a simple random feature scheme (RFS) from prescribed compositional kernels. The compositional kernels we use are inspired by the structure of convolutional…

cs.LG2017★ 17 cited

Depth Separation for Neural Networks

Amit Daniely

Let $f : S^{d - 1} \times S^{d - 1} \to S$ be a function of the form $f (x, x^{'}) = g (⟨ x, x^{'} ⟩)$ for $g:[-1,1]\to \mathb…

cs.LG2017

SGD Learns the Conjugate Kernel Class of the Network

Amit Daniely

We show that the standard stochastic gradient decent (SGD) algorithm is guaranteed to learn, in polynomial time, a function that is competitive with the best function in the conjug…