Pages that link to "Item:Q5154121"
From MaRDI portal
The following pages link to Gradient Descent with Identity Initialization Efficiently Learns Positive-Definite Linear Transformations by Deep Residual Networks (Q5154121):
Displaying 7 items.
- Loss landscapes and optimization in over-parameterized non-linear systems and neural networks (Q2134108) (← links)
- Gradient descent optimizes over-parameterized deep ReLU networks (Q2183586) (← links)
- A survey on deep matrix factorizations (Q2668399) (← links)
- Effects of depth, width, and initialization: A convergence analysis of layer-wise training for deep linear neural networks (Q5037872) (← links)
- (Q5054655) (← links)
- Every Local Minimum Value Is the Global Minimum Value of Induced Model in Nonconvex Machine Learning (Q5214402) (← links)
- Gradient descent for deep matrix factorization: dynamics and implicit bias towards low rank (Q6185686) (← links)
