Deep Layers as Stochastic Solvers

Computer Science

• Optimization Problem 100%
• Single Iteration 100%
• Gradient Method 100%
• Convolutional Network 100%
• Convex Optimization 100%
• Classification Accuracy 100%
• Convolutional Layer 100%

Keyphrases

• Forward Pass 100%
• Dropout 100%
• Multiple Iteration 50%
• Fully Convolutional Layer 50%
• CIFAR-100 50%
• CIFAR-10 50%
• Nonlinear Activation Function 50%
• Dropout Layer 50%
• Proximal Stochastic Gradient 50%
• Linear Layer 50%
• Fully Connected Layer 50%

Mathematics

• Step Size 100%
• Tensor 100%
• Stochastics 100%
• Nonlinear 50%
• Lipschitz Constant 50%
• Connected Layer 50%
• Variance 50%
• Convolutional Layer 50%