Fig 1.
Schematic diagram of multi-round interaction.
Fig 2.
BiLSTM network structure.
Fig 3.
IBLSTM network structure.
Fig 4.
Steps in the execution of translation tasks incorporating soft attention mechanisms.
Fig 5.
Schematic diagram of the network structure of the IABLN recognition algorithm.
Fig 6.
Schematic diagram of the network structure of the IABLN recognition algorithm.
Fig 7.
Experimental steps of modulation pattern recognition method based on IABLN algorithm.
Fig 8.
Comparison of recognition performance using different temporal network networks at different signal-to-noise ratios.
Table 1.
Comparison of indicator evaluation results for different time series networks.
Fig 9.
The recognition accuracy of various signal-to-noise ratios with and without attention mechanism in the network.
Fig 10.
Performance evaluation results of network with and without attention mechanism.
Fig 11.
The weighting of attention mechanism for each output unit when inputting a 14 dB PAM4 signal.
Fig 12.
Modulation pattern recognition results with SNR of -12dB and 0dB in the test set.
Fig 13.
Modulation pattern recognition results with SNR of 6dB and 10dB in the test set.
Fig 14.
Recognition results of IABLN and comparative methods on the test set under different SNRs.
Table 2.
Performance evaluation results of IABLN and comparative algorithms.
Table 3.
Performance evaluation results of the six methods in the CSPB.ML2018 dataset and RML2016.09a dataset.