An optimized hybrid deep learning model using ensemble learning... Fig. 2 Different gait sub-phases event RHC - Right heel contact , LTO - Left toe off, LHC- Left heel contact, ICC- Initial contact of the contralateral limb, RTO — Right toe off,SSL - Swimming limbs opposite to stance limb,VT- Vertical tibia, TS- Terminal swing. Fig.4 Working flowchart of the proposed activities recognition system -ig.5 Proposed hybrid deep learning architecture Fig. 7 Accuracy classification report An optimized hybrid deep learning model using ensemble learning... Table 7 Comparison with state of the art models Accuracy meanwhile takes a constant curve. Figure 9 shows the confusion matrix and accuracy and loss curve using LSTM-—CNN model. Output of CNN-GRU takes almost similar pattern as GRU-CNN, while this model gives us 100% accuracy. We can see the confusion matrix Fig. 10 and the characteristic plot Fig. 11. The ensemble learning model is best suited for the activ- ity recognition problem. Figure 12 shows the best output for ensemble learning model. When evaluating the classification models, it is required to test their accuracy Fig.8 CNN-LSTM model An optimized hybrid deep learning model using ensemble learning... (b) Confusion matrix for LSTM-CNN model. (b) Confusion matrix for CNN-LSTM model. Fig. 10 CNN-GRU model Fig. 12 Ensemble learning model An optimized hybrid deep learning model using ensemble learning... Fig. 13 Individual activity accuracy classification report
