Research Article
Resonant Frequency Modeling of Microwave Antennas Using Gaussian Process Based on Semisupervised Learning
Algorithm 1
Pseudocode of the self-training process.
| (1) | begin | | (2) | e1 ⟵ 1 // assign initial value | | (3) | e2 ⟵ 1 | | (4) | i ⟵ 0 | | (5) | while min (e1, e2) >= error threshold do | | (6) | copy the GP model from the initial training, denoted as GP1 and GP2 | | (7) | i ⟵ i + 1 | | (8) | select num samples Self1.X from the unlabeled sample set | | (9) | input Self1.X into GP1 | | (10) | output Self1.Y | | (11) | select num samples Self2.X from the unlabeled sample set | | (12) | input Self2.X into GP2 | | (13) | output Self2.Y | | (14) | use Self1.GP1 to future train GP2 | | (15) | use Self2.GP2 to future train GP1 | | (16) | use one test sample Test.G by HFSS simulation to test the GP1 and GP2 | | (17) | output e1 and e2 | | (18) | if e1 > e2 | | (19) | then replace GP1 with GP2, denoted as GP | | (20) | else replace GP2 with GP1, denoted as GP | | (21) | end | | (22) | put test sample into existing training sample set to future train GP | | (23) | end | | (24) | end |
|
