Skip to main content
. 2024 Oct 17;10:e2270. doi: 10.7717/peerj-cs.2270

Table 6. The best parameters and performance measures after tuning DNN hyper-parameters.

Dataset Performance metrics before tuning hyper-parameters Best parameter Performance metrics after tuning hyper-parameters
Acc F1 Epochs
(32–500)
Batch size
(32–1,024)
Learning rate
(0.0001–0.1)
Dropout rate
(0.0–0.5)
Acc F1 Pr Rc
ceylon-ide-eclipse 69.86 78.12 496 128 0.002 0.3 72.97 81.86 80.70 83.08
BroadleafCommerce 76.54 85.04 121 32 0.0001 0.5 79.83 87.86 93.22 83.13
hazelcast 61.89 62.96 358 1,024 0.001 0.2 65.19 70.70 66.50 75.72
elasticsearch 68.30 77.57 334 1,024 0.009 0.1 69.45 78.50 83.67 73.93
MapDB 75.34 81.25 381 64 0.0006 0.3 80.82 85.92 87.24 84.65
netty 70.02 78.47 468 1,024 0.008 0.2 76.78 84.71 82.65 86.96
orientdb 61.27 67.33 464 1,024 0.006 0.0 74.50 82.29 81.85 82.95
neo4j 64.44 72.44 343 64 0.0008 0.2 75.39 83.97 87.48 80.83
titan 72.15 81.97 363 1,024 0.008 0.1 77.22 85.37 85.48 85.66
mcMMO 61.76 71.83 390 128 0.006 0.1 66.39 74.67 76.15 73.38
Android-Universal-Image-Loader 60.61 70.45 205 64 0.006 0.4 69.70 77.50 73.91 83.06
antlr4 69.05 79.77 167 512 0.02 0.1 78.57 86.97 83.78 91.05
junit 73.40 82.68 212 1,024 0.007 0.4 80.85 87.86 86.84 89.73
mct 95.45 96.67 76 512 0.0009 0.4 95.45 96.67 93.75 100.00
oryx 85.80 92.05 478 256 0.01 0.3 93.21 96.39 99.32 93.63
Average 71.05 78.57 77.08 84.08 84.16 84.51