Research Article
Semi-supervised Learning for Automatic Modulation Recognition Using Haar Time–Frequency Mask and Positional–Spatial Attention
Table 1
Comparisons between supervised (100%), HTF-PSA-SSL and supervised (1%).
| | Type | Supervised (100%) | HTF-PSA-SSL (%) | Supervised (1%) |
| | 8PSK | 98.00 | 97.00 | 86.00 | | AM-DSB | 80.00 | 78.00 | 66.00 | | AM-SSB | 99.00 | 100.00 | 34.00 | | BPSK | 99.00 | 99.00 | 97.00 | | CPFSK | 100.00 | 100.00 | 100.00 | | GFSK | 100.00 | 100.00 | 100.00 | | PAM4 | 100.00 | 100.00 | 100.00 | | QAM16 | 100.00 | 100.00 | 45.00 | | QAM64 | 99.00 | 99.00 | 65.00 | | QPSK | 98.00 | 98.00 | 97.00 | | WBFM | 50.00 | 42.00 | 47.00 | | Average | 93.00 | 92.09 | 76.09 |
|
|
Note. Bold value indicates the total average result of HTF-PSA-SSL.
|
