Skip to main content

Table 2 Classification performance comparisons on the UCD dataset in the frequency domain

From: A hybrid self-attention deep learning framework for multivariate sleep stage classification

 UCD Dataset (frequency Domain)
MethodAUC-ROCAUC-PRMacro-F1Micro-F1Accuracy
PSVM0.8177 ±0.01420.5767 ±0.01720.5204 ±0.02750.5854 ±0.07330.6193 ±0.1053
DNN0.7213 ±0.14350.5224 ±0.10480.3542 ±0.21710.4331 ±0.22690.5262 ±0.1613
RNN0.6228 ±0.04650.3350 ±0.03940.2663 ±0.02410.3970 ±0.04280.5091 ±0.0391
RNNAtt l0.6172 ±0.03860.3305 ±0.03860.2457 ±0.03070.3734 ±0.05660.5002 ±0.0476
RNNAtt c0.6234 ±0.04510.3335 ±0.03450.2554 ±0.02580.3712 ±0.03250.5010 ±0.0367
CNN0.8732 ±0.01290.6725 ±0.01200.5925 ±0.06040.6492 ±0.08410.6590 ±0.0979
CRNN0.8660 ±0.00740.6454 ±0.01350.5693 ±0.00600.6395 ±0.03700.6634 ±0.0412
CRNNAtt l0.8570 ±0.01830.6281 ±0.03590.5810 ±0.03710.6486 ±0.06410.6683 ±0.0657
CRNNAtt c0.8671 ±0.02740.6418 ±0.04010.5849 ±0.05770.6528 ±0.05470.6791 ±0.0546
ChannelAtt0.8705 ±0.04830.6818 ±0.05800.6517 ±0.03340.7070 ±0.06050.7152 ±0.0574
HybridAtt l0.8719 ±0.02140.6669 ±0.02970.6342 ±0.03160.6962 ±0.06450.7070 ±0.0707
HybridAtt f0.8854 ±0.01370.6886 ±0.02560.6639 ±0.03010.7231 ±0.04890.7328 ±0.0546