Advertisement
Published in:
2017 | OriginalPaper | Chapter
User Selection Based Backpropagation for Siamese Neural Networks in Visual Filters
Authors : Hanju Park, Sukho Lee
Published in: Advanced Multimedia and Ubiquitous Engineering
Publisher: Springer Singapore
Activate our intelligent search to find suitable subject content or patents.
Select sections of text to find matching patents with Artificial Intelligence. powered by
Select sections of text to find additional relevant content using AI-assisted search. powered by
Abstract
In this paper, we propose a user selection based backpropagation method for siamese networks which we will use as visual filters in mobile contents. The loss function used to train the network is affected by the user’s interaction which also affects the update of the weights in the network such that the visual filter takes subjective similarities into account. As a specific application of the proposed algorithm, we expect that the visual filter can be applied for mobile services which provides the user with 3D visual products appearing above the phone. The products appearing in sequence to the user are those which have similar appearances to that selected by the user where the subjective similarity is also taken into account.