In https://hal.archives-ouvertes.fr/hal-01868735 we show how to create deep video representations that match the activity in the human visual system and discover novel relationships between these “deep encodings” and the human visual system.
Presented at CNN 2018 by Hugo Richard.