Zobrazeno 1 - 10
of 26
pro vyhledávání: '"Safran, Itay"'
We study what provable privacy attacks can be shown on trained, 2-layer ReLU neural networks. We explore two types of attacks; data reconstruction attacks, and membership inference attacks. We prove that theoretical results on the implicit bias of 2-
Externí odkaz:
http://arxiv.org/abs/2410.07632
We prove an exponential separation between depth 2 and depth 3 neural networks, when approximating an $\mathcal{O}(1)$-Lipschitz target function to constant accuracy, with respect to a distribution with support in $[0,1]^{d}$, assuming exponentially
Externí odkaz:
http://arxiv.org/abs/2402.07248
We study the size of a neural network needed to approximate the maximum function over $d$ inputs, in the most basic setting of approximating with respect to the $L_2$ norm, for continuous distributions, for a network that uses ReLU activations. We pr
Externí odkaz:
http://arxiv.org/abs/2307.09212
We study the dynamics and implicit bias of gradient flow (GF) on univariate ReLU neural networks with a single hidden layer in a binary classification setting. We show that when the labels are determined by the sign of a target network with $r$ neuro
Externí odkaz:
http://arxiv.org/abs/2205.09072
Autor:
Safran, Itay, Lee, Jason D.
Depth separation results propose a possible theoretical explanation for the benefits of deep neural networks over shallower architectures, establishing that the former possess superior approximation capabilities. However, there are no known results i
Externí odkaz:
http://arxiv.org/abs/2112.02393
Autor:
Safran, Itay, Shamir, Ohad
Recently, there has been much interest in studying the convergence rates of without-replacement SGD, and proving that it is faster than with-replacement SGD in the worst case. However, known lower bounds ignore the problem's geometry, including its c
Externí odkaz:
http://arxiv.org/abs/2106.06880
We study the effects of mild over-parameterization on the optimization landscape of a simple ReLU neural network of the form $\mathbf{x}\mapsto\sum_{i=1}^k\max\{0,\mathbf{w}_i^{\top}\mathbf{x}\}$, in a well-studied teacher-student setting where the t
Externí odkaz:
http://arxiv.org/abs/2006.01005
Autor:
Safran, Itay, Shamir, Ohad
We study the performance of stochastic gradient descent (SGD) on smooth and strongly-convex finite-sum optimization problems. In contrast to the majority of existing theoretical works, which assume that individual functions are sampled with replaceme
Externí odkaz:
http://arxiv.org/abs/1908.00045
Existing depth separation results for constant-depth networks essentially show that certain radial functions in $\mathbb{R}^d$, which can be easily approximated with depth $3$ networks, cannot be approximated by depth $2$ networks, even up to constan
Externí odkaz:
http://arxiv.org/abs/1904.06984
The existence of adversarial examples in which an imperceptible change in the input can fool well trained neural networks was experimentally discovered by Szegedy et al in 2013, who called them "Intriguing properties of neural networks". Since then,
Externí odkaz:
http://arxiv.org/abs/1901.10861