Table 3.
Results of the 3-factor ANOVA evaluating the effects of architecture, sequence length, training method, and first-order interactions on mean classification accuracy. Significant results are displayed in bold.
| Factor | DF | Sum of Squares | F Ratio | Prob > F |
|---|---|---|---|---|
| Architecture | 5 | 0.0090 | 10.0929 | <0.0001 |
| Sequence Length | 1 | 0.0709 | 398.2276 | <0.0001 |
| Training Method | 3 | 0.0021 | 3.9591 | 0.012 |
| Sequence Length*Training Method | 3 | 0.0073 | 13.7292 | <0.0001 |
| Architecture*Sequence Length | 5 | 0.0029 | 3.2372 | 0.0116 |
| Architecture*Training Method | 15 | 0.0039 | 1.4701 | 0.1449 |