Table 2 The optimal values of key parameters
From: A electricity theft detection method through contrastive learning in smart grid
Parameters | Training ratios | |||
|---|---|---|---|---|
50% | 60% | 70% | 80% | |
Learning rate | 0.001 | 0.001 | 0.001 | 0.001 |
Batch size | 32 | 32 | 64 | 64 |
n | 700 | 800 | 800 | 800 |
Discard rate | 10% | 10% | 10% | 10% |
Kernel size | (3,1) | (3,3) | (3,1) | (3,1) |
l | 2 | 2 | 2 | 2 |
\(d_1\) | 1 | 1 | 1 | 1 |
\(d_2\) | 2 | 2 | 2 | 2 |
D | 64 | 64 | 64 | 64 |
d | 16 | 32 | 16 | 16 |
\(\theta\) | 0.95 | 0.9 | 0.95 | 0.9 |
\(\lambda\) | 0.15 | 0.1 | 0.1 | 0.1 |
\(\tau\) | 0.07 | 0.07 | 0.07 | 0.07 |