Table 1 Deep learning models selected hyperparameters used during training

MiniRocket

Number of features: 10000; Maximum dilations per kernel: 16; scoring: MSE

ResNet

Windows size = 24, filter size = 32, kernel sizes: 7, 5 and 4

XceptionTime

Filter size = 16, adaptive average pooling: 32

InceptionTime

Filter size = 32, kernel sizes: 24, depth : 6; dilation: 1

Transformer

Windows size = 24, embedding size: 32, Size of the intermediate

feed forward layer:16, number of layers: 2 and number of heads: 4