Table 3.

Hyper-parameters of models on Ocslab driving dataset.

Model	$\mathbf{Window}\mathbf{Size}\left({\mathit{T}}_{\mathit{x}}\right)$	$\mathbf{Time}\mathbf{Step}\left(\mathsf{\Delta }\mathit{t}\right)$	Selected Features
DeepConvGRU-Attention	90/60/60	45/10/6	original 51D features
DeepConvLSTM-Attention	90/60/60	45/10/6	original 51D features
DeepConvGRU	90/60/60	45/10/6	original 51D features
DeepConvLSTM	90/60/60	45/10/6	original 51D features
CNN	90/60/60	45/10/6	original 51D features
LSTM-51	90/60/60	45/10/6	original 51D features
LSTM-15	90/60/60	45/10/6	selected 15D features
DNN-51	90/60/60	45/10/6	original 51D features
DNN-15	90/60/60	45/10/6	selected 15D features
DNN-45	90/60/60	45/10/6	statistical 45D features
KNN	90/60/60	45/10/6	statistical 45D features
Decision Tree	90/60/60	45/10/6	statistical 45D features
Random Forest	90/60/60	45/10/6	statistical 45D features