We provide visualizations of individual neurons of a deep image recognition network during the temporal process of transfer learning . These visualizations qualitatively demonstrate various novel properties of the transfer learning process regarding the speed and characteristics of adaptation, neuron reuse, and spatial scale of the represented image features . We publish the large-scale dataset that we have created for the purposes of this analysis .

Links: PDF - Abstract

Code :


Keywords : learning - transfer - process - visualizations - scale -

Leave a Reply

Your email address will not be published. Required fields are marked *


Enjoy this blog? Please spread the word :)