Skip To Main Content
Gal Vardi on Muck Rack

Gal Vardi

As seen in: arXiv, Communications of the ACM
Is this you? As a journalist, you can create a free Muck Rack account to customize your profile, list your contact preferences, and upload a portfolio of your best work. Claim your profile
Back to Gal Vardi’s profile

Articles by Gal Vardi

[2505.19087] Temperature is All You Need for Generalization in Langevin Dynamics and other Markov Processes

|
By Gal Vardi, Nathan Srebro, Daniel Soudry, Yonathan Wolanowsky
| arXiv
Work on one of the world's most important websites and make an impact on open science.
Open in Who Shared View Duplicates (372) Issue with article?

[2410.07632] Provable Privacy Attacks on Trained Shallow Neural Networks

|
By Gal Vardi, Itay Safran
| arXiv
Open in Who Shared Issue with article?

[2410.01774] Trained Transformer Classifiers Generalize and Exhibit Benign Overfitting In-Context

|
By Gal Vardi
| arXiv
Open in Who Shared Issue with article?

[2407.15845] Reconstructing Training Data From Real World Models Trained with Transfer Learning

|
By Gilad Yehudai, Gal Vardi, Michal Irani, Itai Antebi
| arXiv
Open in Who Shared Issue with article?

Deconstructing Data Reconstruction: Multiclass, Weight Decay and General Losses

|
By Niv Haim, Gilad Yehudai, Gal Vardi, Yakir Oz
| arXiv
Memorization of training data is an active research area, yet our understanding of the inner workings of neural networks is still in its infancy. Recently, Haim et al. (2022) proposed a scheme to reconstruct training samples from multilayer perceptron binary classifiers, effectively demonstrating that a large portion of training samples are encoded in the parameters of such networks.
Open in Who Shared Issue with article?

[2305.16508] Most Neural Networks Are Almost Learnable

|
By Nathan Srebro, Gal Vardi
| arXiv
We present a PTAS for learning random constant-depth networks. We show that for any fixed $ε>0$ and depth $i$, there is a poly-time algorithm that for any distribution on $\sqrt{d} \cdot \mathbb{S}^{d-1}$ learns random Xavier networks of depth $i$, up to an additive error of $ε$.
Open in Who Shared Issue with article?

On the Implicit Bias in Deep-Learning Algorithms

|
By Gal Vardi
| Communications of the ACM Verified
By Gal Vardi Communications of the ACM, June 2023, Vol. 66 No. 6, Pages 86-93 10.1145/3571070 Comments View as: Print Mobile App ACM Digital Library In the Digital Edition Share: Send by email Share on reddit Share on StumbleUpon Share on Hacker News Share on Tweeter Share on Facebook Deep learning has been highly successful in recent years and has led to dramatic improvements in multiple domains.
Open in Who Shared Issue with article?

[2305.03350] Reconstructing Training Data from Multiclass Neural Networks

|
By Gilad Yehudai, Niv Haim, Gal Vardi, Michal Irani
| arXiv
Reconstructing samples from the training set of trained neural networks is a major privacy concern. Haim et al. (2022) recently showed that it is possible to reconstruct training samples from neural network binary classifiers, based on theoretical results about the implicit bias of gradient methods.
Open in Who Shared Issue with article?
Show More