Figure 3 - data

Figure 3. Influence of number of channels and analysis window on training performance. a) Final validation F1-score of all trained models depending on the number of input channels: one (pyramidal channel; see methods), three (pyramidal channel and extreme channels), or eight (all channels of the probe). b) Same as panel a), but depending on the time window used for analysis.

图3 通道数量与分析窗口对训练性能的影响。a) 所有已训练模型的最终验证集F1分数（F1-score）随输入通道数的变化情况：1个通道（锥状通道（pyramidal channel）；详见方法部分）、3个通道（锥状通道与极值通道（extreme channels）），或8个通道（探头（probe）的全部通道）。b) 与子图a)一致，但横轴为分析所用的时间窗口。