Picture for Jianhao Ma

Jianhao Ma

Convergence of Gradient Descent with Small Initialization for Unregularized Matrix Completion

Add code
Feb 09, 2024
Viaarxiv icon

Robust Sparse Mean Estimation via Incremental Learning

Add code
May 24, 2023
Viaarxiv icon

On the Optimization Landscape of Burer-Monteiro Factorization: When do Global Solutions Correspond to Ground Truth?

Add code
Feb 21, 2023
Viaarxiv icon

Behind the Scenes of Gradient Descent: A Trajectory Analysis via Basis Function Decomposition

Add code
Oct 04, 2022
Figure 1 for Behind the Scenes of Gradient Descent: A Trajectory Analysis via Basis Function Decomposition
Figure 2 for Behind the Scenes of Gradient Descent: A Trajectory Analysis via Basis Function Decomposition
Figure 3 for Behind the Scenes of Gradient Descent: A Trajectory Analysis via Basis Function Decomposition
Figure 4 for Behind the Scenes of Gradient Descent: A Trajectory Analysis via Basis Function Decomposition
Viaarxiv icon

Blessing of Nonconvexity in Deep Linear Models: Depth Flattens the Optimization Landscape Around the True Solution

Add code
Jul 15, 2022
Figure 1 for Blessing of Nonconvexity in Deep Linear Models: Depth Flattens the Optimization Landscape Around the True Solution
Figure 2 for Blessing of Nonconvexity in Deep Linear Models: Depth Flattens the Optimization Landscape Around the True Solution
Figure 3 for Blessing of Nonconvexity in Deep Linear Models: Depth Flattens the Optimization Landscape Around the True Solution
Figure 4 for Blessing of Nonconvexity in Deep Linear Models: Depth Flattens the Optimization Landscape Around the True Solution
Viaarxiv icon

Global Convergence of Sub-gradient Method for Robust Matrix Recovery: Small Initialization, Noisy Measurements, and Over-parameterization

Add code
Feb 17, 2022
Figure 1 for Global Convergence of Sub-gradient Method for Robust Matrix Recovery: Small Initialization, Noisy Measurements, and Over-parameterization
Figure 2 for Global Convergence of Sub-gradient Method for Robust Matrix Recovery: Small Initialization, Noisy Measurements, and Over-parameterization
Figure 3 for Global Convergence of Sub-gradient Method for Robust Matrix Recovery: Small Initialization, Noisy Measurements, and Over-parameterization
Figure 4 for Global Convergence of Sub-gradient Method for Robust Matrix Recovery: Small Initialization, Noisy Measurements, and Over-parameterization
Viaarxiv icon

Towards Understanding Generalization via Decomposing Excess Risk Dynamics

Add code
Jun 11, 2021
Figure 1 for Towards Understanding Generalization via Decomposing Excess Risk Dynamics
Figure 2 for Towards Understanding Generalization via Decomposing Excess Risk Dynamics
Viaarxiv icon

Implicit Regularization of Sub-Gradient Method in Robust Matrix Recovery: Don't be Afraid of Outliers

Add code
Feb 05, 2021
Figure 1 for Implicit Regularization of Sub-Gradient Method in Robust Matrix Recovery: Don't be Afraid of Outliers
Figure 2 for Implicit Regularization of Sub-Gradient Method in Robust Matrix Recovery: Don't be Afraid of Outliers
Figure 3 for Implicit Regularization of Sub-Gradient Method in Robust Matrix Recovery: Don't be Afraid of Outliers
Figure 4 for Implicit Regularization of Sub-Gradient Method in Robust Matrix Recovery: Don't be Afraid of Outliers
Viaarxiv icon