Scalable kernel methods via doubly stochastic gradients

Keyphrases

• Kernel Methods 100%
• Doubly Stochastic 100%
• Stochastic Gradient 100%
• Neural Nets 50%
• Functional Gradient 50%
• Random Features 33%
• Learning Problems 16%
• Optimization Problem 16%
• Convex Optimization Problem 16%
• Class Function 16%
• Generalization Bounds 16%
• ImageNet 16%
• Competitive Performance 16%
• Energy Materials 16%
• Stochastic Approximation 16%
• General Perception 16%
• Optimal Function 16%
• Training Points 16%
• Reproducing Kernel Hilbert Space 16%
• Handwritten Digits 16%
• MNIST 16%
• Random Training 16%
• Convolution Feature 16%
• Nonlinear Learning 16%

Mathematics

• Stochastics 100%
• Kernel Method 100%
• Neural Net 50%
• Hilbert Space 16%
• Convolution 16%

Computer Science

• Kernel Method 100%
• Optimization Problem 33%
• Convex Optimization 16%
• Reproducing Kernel 16%
• Learning Problem 16%
• Stochastic Approximation 16%
• Hilbert Space 16%
• Iteration Converges 16%
• Training Point 16%
• handwritten digit 16%