Kernel learning on LSK synthetic data 1.0 (test)
77.7Relative ErrorRegular Fourier Features (real)
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| Regular Fourier Features (real)network output=real-valued, features (m)=255, cutoff frequency=10, rank (r)=8, hidden layers=2, units=1282026.02 | 77.7 | 1.95 | 6.5 | |
| Regular Fourier Features (complex)network output=complex-valued, features (m)=255, cutoff frequency=10, rank (r)=8, hidden layers=2, units=1282026.02 | 81.5 | 1.85 | 8.5 | |
| RBF2026.02 | 94.2 | 1.94 | 6.3 | |
| NGSMhidden layers=2, units=1282026.02 | 96.1 | 1.84 | 8.4 | |
| NNK2026.02 | 148.2 | 1.94 | 5.7 | |
| DKLhidden layers=2, units=1282026.02 | 373.5 | 0.4 | 30.4 |