I have a dataset of some activities. The dataset contains the status of different sensors and the label of activity. T trained a model in Keras with the following architecture which models the activities.
Now, I need to make a different LSTM-based model for each activity separately. Then, calculate the probability of creating the sequence by each model. Can I do that with Keras? How should I design the architecture of the model? Is it right to remove the dense layer and then create a separate model for each activity label separately?