Table 5.
Sensitivity analysis for H-PABO single objective optimization.
| HPs | Experiment 1 | Experiment 2 | Experiment 3 | Experiment 4 | |
|---|---|---|---|---|---|
| Input encoding HPs | bk | 2 | 2 | 2 | 2 |
| pk | 8 | 12 | 8 | 8 | |
| Charge | [0, 0.5] | [0, 0.5] | [0, 0.5] | [0, 0.5] | |
| Function | Flip-flop | Flip-flop | Flip-flop | Flip-flop | |
| Interval | 1 | 5 | 1 | 2 | |
| EONS HPs | Population size | 1000 | 1500 | 400 | 1000 |
| Mutation rate | 0.9 | 0.9 | 0.9 | 0.9 | |
| Crossover rate | 0.5 | 0.4 | 0.5 | 0.7 | |
| Accelerator HPs | Synp weight | [−255, 255] | [−127, 127] | [−255, 255] | - |
| Neuron threshold | [0, 1023] | [0, 1023] | [0, 1023] | - | |
| Synp delay | 127 | 255 | 15 | ||
| Neuromorphic System Performance | 52% | 70.99% | 50% | 53% | |
| Accelerator | DANNA2 | DANNA2 | DANNA2 | mrDANNA | |
| Application | Pole-Balance | Pole-Balance | RoboNav | RoboNav | |