Research Article
Probe Selection and Power Weighting in Multiprobe OTA Testing: A Neural Network-Based Approach
Algorithm 1
Training process for the proposed NN model.
| Input: Training data set | | Output: Weight vector | | (1) | Initialize: the iteration index , , , λ, η, γ. | | (2) | while do | | (3) | Randomly select a mini-batch of size S from ; | | (4) | Calculate the gradient and the momentum according to (9) and (10), respectively; | | (5) | Update according to (11); | | (6) | Adjust : , and ; | | (7) | if then | | (8) | if then | | (9) | | | (10) | else if then | | (11) | | | (12) | else if | | (13) | end if | | (14) | Update . | | (15) | end while |
|
