Research Article
Hybridizing Evolutionary Computation and Deep Neural Networks: An Approach to Handwriting Recognition Using Committees and Transfer Learning
Table 4
Summary of accuracies of the best 20 GE individuals after full training in EMNIST Letters.
| # | Mean | Std. Dev. | Min. | Median | Max. |
| 1 | 94.6585 | 0.074004 | 94.665 | 94.51 | 94.76 | 2 | 94.7300 | 0.070934 | 94.740 | 94.61 | 94.86 | 3 | 94.5635 | 0.084870 | 94.580 | 94.41 | 94.70 | 4 | 95.1215 | 0.049553 | 95.120 | 95.02 | 95.19 | 5 | 94.2790 | 0.064880 | 94.275 | 94.16 | 94.38 | 6 | 94.7230 | 0.068832 | 94.710 | 94.63 | 94.88 | 7 | 94.6540 | 0.053646 | 94.650 | 94.53 | 94.74 | 8 | 93.7270 | 0.094429 | 93.705 | 93.60 | 93.93 | 9 | 93.6515 | 0.070058 | 93.645 | 93.50 | 93.75 | 10 | 94.5305 | 0.139075 | 94.550 | 94.18 | 94.75 | 11 | 94.6890 | 0.045410 | 94.690 | 94.61 | 94.78 | 12 | 94.4335 | 0.071545 | 94.440 | 94.28 | 94.55 | 13 | 94.4330 | 0.075888 | 94.435 | 94.28 | 94.61 | 14 | 94.6260 | 0.132045 | 94.630 | 94.24 | 94.81 | 15 | 94.4605 | 0.062616 | 94.465 | 94.35 | 94.57 | 16 | 94.1590 | 0.093578 | 94.165 | 94.00 | 94.29 | 17 | 93.4505 | 0.156423 | 93.455 | 93.14 | 93.81 | 18 | 93.5095 | 0.149929 | 93.505 | 93.25 | 93.77 | 19 | 94.2725 | 0.074684 | 94.270 | 94.13 | 94.38 | 20 | 94.5085 | 0.070208 | 94.495 | 94.41 | 94.64 |
|
|