# How to calculate the attention loss in the paper “Tell Me Where to Look: Guided Attention Inference Network”?

I have been reading the research paper Tell Me Where to Look: Guided Attention Inference Network.

In this paper, they calculate the attention loss, but I didn't understand how to calculate it. Do we have to calculate it like outcome[c]? If it is so, then why do arrows connect with each other from the middle FC and the last FC?

Here is the image:

• From this figure, there are apparently 3 FC, and the second (or middle) FC does not seem to be connected to the last FC. So, can you clarify what you mean by "why do arrows connect with each other from the middle FC and the last FC?"? – nbro Aug 21 '20 at 11:46
• FC (where the red box is turned on) layers connected by green attention mining loss. – dato nefaridze Aug 21 '20 at 11:58