Table 3 Comparison of OSS-CNN with digital neural network architectures in terms of accuracy and trainable parameters.
From: Photonic neuromorphic accelerators for event-based imaging flow cytometry
Architectures | Max accuracy (%) | Mean accuracy (%) | Hiddenunits | Parameters |
|---|---|---|---|---|
1 FCL | 97.5 | 96.8 ± 0.7 | – | 30,003 |
2 FCLs | 98.2 | 97.9 ± 0.3 | 46 | 460,187 |
OSS-1FCL | 97.9 | 97.6 ± 0.3 | – | 1947 |
OSS-2FCL | 98.6 | 97.8 ± 0.6 | 80 | 52,163 |
GRU-RNN | 98.6 | 97.9 ± 0.7 | 38 | 1,144,677 |
