Table 6.

The classification results for different models with offline and on-device training are compared. We show three different models (CNN-LSTM, ResNet, and DeepConvLSTM) with both offline and on-device training for the OPPORTUNITY dataset. The best scores are emboldened row-wise per metric.

Model	Offline Training		On-Device Training
	Accuracy	F1	Accuracy	F1
CNN-LSTM	0.644 $(\pm 0.08)$	0.612 $(\pm 0.09)$	0.745 $(\pm 0.09)$	0.703 $(\pm 0.07)$
ResNet	0.493 $(\pm 0.12)$	0.487 $(\pm 0.11)$	0.727 $(\pm 0.10)$	0.671 $(\pm 0.12)$
DeepConvLSTM	0.598 $(\pm 0.09)$	0.599 $(\pm 0.09)$	0.654 $(\pm 0.07)$	0.615 $(\pm 0.06)$