Table 5 Accuracy on QUIC dataset.
From: Method for multi-task learning fusion network traffic classification to address small sample labels
Accuracy [bandwidth, duration, traffic class] | ||||
|---|---|---|---|---|
Size of labeled samples | RF36 | CNN34 | CNN+RNN25 | CNN-multi-task learning |
10 | [–,–,48.67%] | [–,–,85.33%] | [89.33%, 92.00%, 64.67%] | [91.89%,74.00%,90.00%] |
20 | [–,–,64.00%] | [–,–,87.33%] | [89.33%, 92.00%, 66.67%] | [92.89%,74.33%,91.00%] |
50 | [–,–,78.00%] | [–,–,88.83%] | [89.33%, 92.00%, 76.67%] | [93.33%,74.33%,92.87%] |
150 | [–,–,86.67%] | [–,–,90.00%] | [89.33%, 92.00%, 85.33%] | [93.33%,75.33%,94.00%] |
