Table 7 Test accuracy of XLG-Net on the text classification datasets.
Mixup layer set | MR | \(\Delta _{MR}\) | Ohsumed | \(\Delta _{Ohsumed}\) | R52 | \(\Delta _{R52}\) | R8 | \(\Delta _{R8}\) |
|---|---|---|---|---|---|---|---|---|
\(\varnothing\) | 87.03 | 0 | 72.66 | 0 | 96.44 | 0 | 98.29 | 0 |
{1} | 88.29 | +1.26 | 72.92 | +0.26 | 96.46 | +0.02 | 98.33 | +0.04 |
{1,2} | 88.07 | +1.04 | 72.70 | +0.04 | 96.44 | 0 | 98.31 | +0.02 |
{3,4,5} | 88.33 | +1.30 | 72.97 | +0.31 | 96.46 | +0.02 | 98.33 | +0.04 |
{5,6} | 88.16 | +1.13 | 72.75 | +0.09 | 96.45 | +0.01 | 98.32 | +0.03 |
{3,6,9} | 88.35 | +1.32 | 73.21 | +0.55 | 96.48 | +0.04 | 98.38 | +0.09 |
{7,8,9} | 88.31 | +1.28 | 73.19 | +0.53 | 96.48 | +0.04 | 98.36 | +0.07 |
{6,7,9,12} | 88.29 | +1.26 | 73.12 | +0.46 | 96.47 | +0.03 | 98.34 | +0.05 |
{9,10,12} | 88.57 | +1.54 | 73.34 | +0.68 | 96.50 | +0.06 | 98.41 | +0.12 |
{3,4,6,7,9,12} | 88.22 | +1.19 | 73.07 | +0.41 | 96.46 | +0.02 | 98.33 | +0.04 |
