Loss landscapes and optimization in over-parameterized non-linear systems and neural networks (Q2134108): Difference between revisions

@@ Property / cites work @@
+Gradient Descent with Identity Initialization Efficiently Learns Positive-Definite Linear Transformations by Deep Residual Networks
+Normal rank
@@ Property / cites work @@
+Reconciling modern machine-learning practice and the classical bias–variance trade-off
+Normal rank
@@ Property / cites work @@
+Condition
@@ Property / cites work: Condition / rank @@
+Normal rank
@@ Property / cites work @@
+Condition Numbers of Gaussian Random Matrices
@@ Property / cites work: Condition Numbers of Gaussian Random Matrices / rank @@
+Normal rank
@@ Property / cites work @@
+Global Minima of Overparameterized Neural Networks
+Normal rank
@@ Property / cites work @@
+Q4998970
@@ Property / cites work: Q4998970 / rank @@
+Normal rank
@@ Property / cites work @@
+Wide neural networks of any depth evolve as linear models under gradient descent                   <sup>*</sup>
+Normal rank
@@ Property / cites work @@
+Q5643297
@@ Property / cites work: Q5643297 / rank @@
+Normal rank
@@ Property / cites work @@
+A mean field view of the landscape of two-layer neural networks
+Normal rank
@@ Property / cites work @@
+Q5491447
@@ Property / cites work: Q5491447 / rank @@
+Normal rank
@@ Property / cites work @@
+Q5593503
@@ Property / cites work: Q5593503 / rank @@
+Normal rank
@@ Property / cites work @@
+Theoretical Insights Into the Optimization Landscape of Over-Parameterized Shallow Neural Networks
+Normal rank
@@ Property / cites work @@
+A jamming transition from under- to over-parametrization affects generalization in deep learning
+Normal rank
@@ Property / cites work @@
+Gradient descent optimizes over-parameterized deep ReLU networks
+Normal rank