Nyström method vs random Fourier features: A theoretical and empirical comparison

Keyphrases

• Random Fourier Features 100%
• Large Gap 25%
• Data Dependency 25%
• Matrix Approach 25%
• Feature-based 25%
• Eigenvalue Spectrum 25%
• Training Data 25%
• Cosine Function 25%
• Sine Function 25%
• Generalization Performance 25%
• Kernel Matrix 25%
• Training Examples 25%
• Kernel Learning 25%
• Independent Distribution 25%
• Generalization Error Bound 25%

Computer Science

• Basis Function 100%
• Large Data Set 50%
• Training Data 50%
• Generalization Performance 50%
• Training Example 50%
• Generalization Error 50%
• Matrix Approach 50%