ERD

On the Optimal Memorization Power of ReLU Neural Networks

Project: THUNDEEP

Updated at: 31-08-2024

Show more information

Project acronym THUNDEEP

On Margin Maximization in Linear and ReLU Networks

Project: THUNDEEP

Updated at: 31-08-2024

Show more information

Project acronym THUNDEEP

The Effects of Mild Over-parameterization on the Optimization Landscapeof Shallow ReLU Neural Networks

Project: THUNDEEP

Updated at: 27-04-2024

Show more information

Project acronym THUNDEEP

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

Project: THUNDEEP

Updated at: 27-04-2024

Show more information

Project acronym THUNDEEP

On the Power and Limitations of Random Features for Understanding Neural Networks

Project: THUNDEEP

Updated at: 27-04-2024

Show more information

Project acronym THUNDEEP

Exponential Convergence Time of Gradient Descent for One-Dimensional Deep Linear Neural Networks

Project: THUNDEEP

Updated at: 27-04-2024

Show more information

Project acronym THUNDEEP

Depth Separations in Neural Networks: What is Actually Being Separated?

Project: THUNDEEP

Updated at: 27-04-2024

Show more information

Project acronym THUNDEEP

Implicit Regularization in ReLU Networks with the Square Loss

Project: THUNDEEP

Updated at: 27-04-2024

Show more information

Project acronym THUNDEEP

Oracle Complexity in Nonsmooth Nonconvex Optimization

Project: THUNDEEP

Updated at: 27-04-2024

Show more information

Project acronym THUNDEEP

Are ResNets Provably Better than Linear Predictors?

Project: THUNDEEP

Updated at: 27-04-2024

Show more information

Project acronym THUNDEEP

The Implicit Bias of Benign Overfitting

Project: THUNDEEP

Updated at: 27-04-2024

Show more information

Project acronym THUNDEEP

Reconstructing Training Data from Trained Neural Networks

Project: THUNDEEP

Updated at: 27-04-2024

Show more information

Project acronym THUNDEEP

Gradient Methods Provably Converge to Non-Robust Networks

Project: THUNDEEP

Updated at: 27-04-2024

Show more information

Project acronym THUNDEEP

The Complexity of Making the Gradient Small in Stochastic Convex Optimization

Project: THUNDEEP

Updated at: 27-04-2024

Show more information

Project acronym THUNDEEP

Size and Depth Separation in Approximating Natural Functions with Neural Networks

Project: THUNDEEP

Updated at: 27-04-2024

Show more information

Project acronym THUNDEEP

Proving the Lottery Ticket Hypothesis: Pruning is All You Need

Project: THUNDEEP

Updated at: 27-04-2024

Show more information

Project acronym THUNDEEP

Random Shuffling Beats SGD Only After Many Epochs on Ill-Conditioned Problems

Project: THUNDEEP

Updated at: 27-04-2024

Show more information

Project acronym THUNDEEP

Can We Find Near-Approximately-Stationary Points of Nonsmooth Nonconvex Functions?

Project: THUNDEEP

Updated at: 27-04-2024

Show more information

Project acronym THUNDEEP

On the Complexity of Finding Small Subgradients in Nonsmooth Optimization

Project: THUNDEEP

Updated at: 27-04-2024

Show more information

Project acronym THUNDEEP

Learning a Single Neuron with Gradient Methods

Project: THUNDEEP

Updated at: 27-04-2024

Show more information

Project acronym THUNDEEP

How Good is SGD with Random Shuffling?

Project: THUNDEEP

Updated at: 27-04-2024

Show more information

Project acronym THUNDEEP

Gradient Methods Never Overfit On Separable Data

Project: THUNDEEP

Updated at: 27-04-2024

Show more information

Project acronym THUNDEEP

Learning a Single Neuron with Bias Using Gradient Descent

Project: THUNDEEP

Updated at: 27-04-2024

Show more information

Project acronym THUNDEEP

The Sample Complexity of One-Hidden-Layer Neural Networks

Project: THUNDEEP

Updated at: 27-04-2024

Show more information

Project acronym THUNDEEP

Adversarial Examples Exist in Two-Layer ReLU Networks for Low Dimensional Data Manifolds

Project: THUNDEEP

Updated at: 27-04-2024

Show more information

Project acronym THUNDEEP

Width is Less Important than Depth in ReLU Neural Networks

Project: THUNDEEP

Updated at: 27-04-2024

Show more information

Project acronym THUNDEEP

Implicit Regularization Towards Rank Minimization in ReLU Networks

Project: THUNDEEP

Updated at: 27-04-2024

Show more information

Project acronym THUNDEEP

Neural Networks with Small Weights and Depth-Separation Barriers

Project: THUNDEEP

Updated at: 27-04-2024

Show more information

Project acronym THUNDEEP

Periodic Reporting for period 3 - THUNDEEP (A Theory for Understanding, Designing, and Training Deep Learning Systems)

Project: THUNDEEP

Updated at: 27-04-2024

Show more information

Project acronym THUNDEEP

Resources

Selected filters:

On the Optimal Memorization Power of ReLU Neural Networks

On Margin Maximization in Linear and ReLU Networks

The Effects of Mild Over-parameterization on the Optimization Landscapeof Shallow ReLU Neural Networks

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

On the Power and Limitations of Random Features for Understanding Neural Networks

Exponential Convergence Time of Gradient Descent for One-Dimensional Deep Linear Neural Networks

Depth Separations in Neural Networks: What is Actually Being Separated?

Implicit Regularization in ReLU Networks with the Square Loss

Oracle Complexity in Nonsmooth Nonconvex Optimization

Are ResNets Provably Better than Linear Predictors?

The Implicit Bias of Benign Overfitting

Reconstructing Training Data from Trained Neural Networks

Gradient Methods Provably Converge to Non-Robust Networks

The Complexity of Making the Gradient Small in Stochastic Convex Optimization

Size and Depth Separation in Approximating Natural Functions with Neural Networks

Proving the Lottery Ticket Hypothesis: Pruning is All You Need

Random Shuffling Beats SGD Only After Many Epochs on Ill-Conditioned Problems

Can We Find Near-Approximately-Stationary Points of Nonsmooth Nonconvex Functions?

On the Complexity of Finding Small Subgradients in Nonsmooth Optimization

Learning a Single Neuron with Gradient Methods

How Good is SGD with Random Shuffling?

Gradient Methods Never Overfit On Separable Data

Learning a Single Neuron with Bias Using Gradient Descent

The Sample Complexity of One-Hidden-Layer Neural Networks

Adversarial Examples Exist in Two-Layer ReLU Networks for Low Dimensional Data Manifolds

Width is Less Important than Depth in ReLU Neural Networks

Implicit Regularization Towards Rank Minimization in ReLU Networks

Neural Networks with Small Weights and Depth-Separation Barriers

Periodic Reporting for period 3 - THUNDEEP (A Theory for Understanding, Designing, and Training Deep Learning Systems)