How does momentum benefit deep neural networks architecture design? A few case studies (Q2079522): Difference between revisions

@@ Property / describes a project that uses @@
+Longformer
@@ Property / describes a project that uses: Longformer / rank @@
+Normal rank
@@ Property / describes a project that uses @@
+Synthesizer
@@ Property / describes a project that uses: Synthesizer / rank @@
+Normal rank
@@ Property / describes a project that uses @@
+GitHub
@@ Property / describes a project that uses: GitHub / rank @@
+Normal rank
@@ Property / describes a project that uses @@
+PixelCNN++
@@ Property / describes a project that uses: PixelCNN++ / rank @@
+Normal rank
@@ Property / describes a project that uses @@
+Tensor2Tensor
@@ Property / describes a project that uses: Tensor2Tensor / rank @@
+Normal rank
@@ Property / describes a project that uses @@
+MuJoCo
@@ Property / describes a project that uses: MuJoCo / rank @@
+Normal rank
@@ Property / describes a project that uses @@
+MNIST
@@ Property / describes a project that uses: MNIST / rank @@
+Normal rank
@@ Property / describes a project that uses @@
+Adam
@@ Property / describes a project that uses: Adam / rank @@
+Normal rank
@@ Property / describes a project that uses @@
+BLEU
@@ Property / describes a project that uses: BLEU / rank @@
+Normal rank
@@ Property / describes a project that uses @@
+ViT
@@ Property / describes a project that uses: ViT / rank @@
+Normal rank
@@ Property / describes a project that uses @@
+Nyströmformer
@@ Property / describes a project that uses: Nyströmformer / rank @@
+Normal rank
@@ Property / MaRDI profile type @@
+MaRDI publication profile
@@ Property / MaRDI profile type: MaRDI publication profile / rank @@
+Normal rank
@@ Property / OpenAlex ID @@
+W3205535720
@@ Property / OpenAlex ID: W3205535720 / rank @@
+Normal rank
@@ Property / cites work @@
+THE HEAVY BALL WITH FRICTION METHOD, I. THE CONTINUOUS DYNAMICAL SYSTEM: GLOBAL EXPLORATION OF THE LOCAL MINIMA OF A REAL-VALUED FUNCTION BY ASYMPTOTIC ANALYSIS OF A DISSIPATIVE DYNAMICAL SYSTEM
+Normal rank
@@ Property / cites work @@
+Stable architectures for deep neural networks
@@ Property / cites work: Stable architectures for deep neural networks / rank @@
+Normal rank
@@ Property / cites work @@
+Q3320132
@@ Property / cites work: Q3320132 / rank @@
+Normal rank
@@ Property / cites work @@
+Some methods of speeding up the convergence of iteration methods
+Normal rank
@@ Property / cites work @@
+Adversarial defense via the data-dependent activation, total variation minimization, and adversarial training
+Normal rank
@@ Property / cites work @@
+Scheduled Restart Momentum for Accelerated Stochastic Gradient Descent
+Normal rank
@@ Property / cites work @@
+Graph interpolating activation improves both natural and robust accuracies in data-efficient deep learning
+Normal rank
@@ Property / cites work @@
+A variational perspective on accelerated methods in optimization
+Normal rank