Picture for Alon Brutzkus

Alon Brutzkus

How Uniform Random Weights Induce Non-uniform Bias: Typical Interpolating Neural Networks Generalize with Narrow Teachers

Feb 09, 2024
Viaarxiv icon

A Theoretical Analysis of Fine-tuning with Linear Teachers

Jul 04, 2021
Figure 1 for A Theoretical Analysis of Fine-tuning with Linear Teachers
Figure 2 for A Theoretical Analysis of Fine-tuning with Linear Teachers
Figure 3 for A Theoretical Analysis of Fine-tuning with Linear Teachers
Viaarxiv icon

Towards Understanding Learning in Neural Networks with Linear Teachers

Jan 07, 2021
Figure 1 for Towards Understanding Learning in Neural Networks with Linear Teachers
Figure 2 for Towards Understanding Learning in Neural Networks with Linear Teachers
Figure 3 for Towards Understanding Learning in Neural Networks with Linear Teachers
Figure 4 for Towards Understanding Learning in Neural Networks with Linear Teachers
Viaarxiv icon

On the Inductive Bias of a CNN for Orthogonal Patterns Distributions

Feb 22, 2020
Figure 1 for On the Inductive Bias of a CNN for Orthogonal Patterns Distributions
Figure 2 for On the Inductive Bias of a CNN for Orthogonal Patterns Distributions
Figure 3 for On the Inductive Bias of a CNN for Orthogonal Patterns Distributions
Viaarxiv icon

On the Optimality of Trees Generated by ID3

Jul 11, 2019
Figure 1 for On the Optimality of Trees Generated by ID3
Figure 2 for On the Optimality of Trees Generated by ID3
Figure 3 for On the Optimality of Trees Generated by ID3
Viaarxiv icon

ID3 Learns Juntas for Smoothed Product Distributions

Jun 20, 2019
Viaarxiv icon

Low Latency Privacy Preserving Inference

Add code
Dec 27, 2018
Figure 1 for Low Latency Privacy Preserving Inference
Figure 2 for Low Latency Privacy Preserving Inference
Figure 3 for Low Latency Privacy Preserving Inference
Figure 4 for Low Latency Privacy Preserving Inference
Viaarxiv icon

SGD Learns Over-parameterized Networks that Provably Generalize on Linearly Separable Data

Oct 27, 2017
Figure 1 for SGD Learns Over-parameterized Networks that Provably Generalize on Linearly Separable Data
Viaarxiv icon

Globally Optimal Gradient Descent for a ConvNet with Gaussian Inputs

Feb 26, 2017
Figure 1 for Globally Optimal Gradient Descent for a ConvNet with Gaussian Inputs
Figure 2 for Globally Optimal Gradient Descent for a ConvNet with Gaussian Inputs
Figure 3 for Globally Optimal Gradient Descent for a ConvNet with Gaussian Inputs
Figure 4 for Globally Optimal Gradient Descent for a ConvNet with Gaussian Inputs
Viaarxiv icon