Table 3.
Parameter | Description | Assignments |
---|---|---|
K | No. of context vectors | 2, 4 |
[ϵb, ϵn] | Learning rates | [0.1, 0.001], [0.3, 0.003], [0.5, 0.005] |
Activation thresholds | [0.1, 0.01], [0.3, 0.03], [0.5, 0.05], [0.01, 0.01], [0.05, 0.05], [0.07, 0.07] |
The best results averaged across all grid search iterations are shown in bold letters. For further descriptions of the parameters, see Table 1.