Table 3 Comparison on computational complexity between PCA methods and Hebbian eigenfilter
From: Stream-based Hebbian eigenfilter for real-time neuronal spike discrimination
| Â | Â | #Add. | #Multi. | #Div. | #Square root | #Memory |
|---|---|---|---|---|---|---|
|  |  | × 10 6 | × 10 6 | × 10 3 | ×10 3 | (bits) × 10 6 |
Hebbian eigenfilter (HE) | 0.46 | 0.59 | 0 | 0 | 1.05 | |
Stream-based HE (SHE) | 0.46 | 0.59 | 0 | 0 | 0.0031 | |
Orthogonal iteration (OI) based PCA | 4.51 | 4.39 | 0.016 | 0.016 | 1.18 | |
QR method based PCA | 4.77 | 4.75 | 1.1 | 0.57 | 1.25 | |
Jacobi method based PCA | 10.6 | 8.94 | 48.4 | 322 | 1.18 | |
Computational | OI vs SHE | 89.8% | 86.6% | 100% | 100% | 99.7% |
complexity improvement | QR vs SHE | 90.4% | 87.6% | 100% | 100% | 99.8% |
(reduction rate) | Jacobi vs SHE | 95.7% | 93.4% | 100% | 100% | 99.7% |