Picture for Gilad Yehudai

Gilad Yehudai

From Tempered to Benign Overfitting in ReLU Neural Networks

May 24, 2023
Guy Kornowski, Gilad Yehudai, Ohad Shamir

* 43 pages

Access Paper or Ask Questions

Reconstructing Training Data from Multiclass Neural Networks

May 05, 2023
Gon Buzaglo, Niv Haim, Gilad Yehudai, Gal Vardi, Michal Irani

Access Paper or Ask Questions

Gradient Methods Provably Converge to Non-Robust Networks

Feb 09, 2022
Gal Vardi, Gilad Yehudai, Ohad Shamir

Figure 1 for Gradient Methods Provably Converge to Non-Robust Networks

Figure 2 for Gradient Methods Provably Converge to Non-Robust Networks

Access Paper or Ask Questions

Width is Less Important than Depth in ReLU Neural Networks

Feb 08, 2022
Gal Vardi, Gilad Yehudai, Ohad Shamir

Access Paper or Ask Questions

On the Optimal Memorization Power of ReLU Neural Networks

Oct 07, 2021
Gal Vardi, Gilad Yehudai, Ohad Shamir

Access Paper or Ask Questions

Learning a Single Neuron with Bias Using Gradient Descent

Jun 02, 2021
Gal Vardi, Gilad Yehudai, Ohad Shamir

Figure 1 for Learning a Single Neuron with Bias Using Gradient Descent

Figure 2 for Learning a Single Neuron with Bias Using Gradient Descent

Access Paper or Ask Questions

The Connection Between Approximation, Depth Separation and Learnability in Neural Networks

Jan 31, 2021
Eran Malach, Gilad Yehudai, Shai Shalev-Shwartz, Ohad Shamir

Access Paper or Ask Questions