Skip to main content
. Author manuscript; available in PMC: 2020 Oct 19.
Published in final edited form as: IEEE EMBS Int Conf Biomed Health Inform. 2019 Sep 12;2019:10.1109/BHI.2019.8834637. doi: 10.1109/BHI.2019.8834637

Figure 3:

Figure 3:

The structure of the attention mechanism used in the proposed model. At each time step t, the attention module utilizes Ft and the previous hidden state of the RNN part, ht-1 to calculate an expected value, ft with respect to vertical feature slices, Ft,I and the importance of each region of input window frame, αt,i