Sorry, you need to enable JavaScript to visit this website.

facebooktwittermailshare

Deep Metric Learning using Similarities from 
 Nonlinear Rank Approximations

Abstract: 

In recent years, deep metric learning has achieved promising results in learning high dimensional semantic feature embeddings where the spatial relationships of the feature vectors match the visual similarities of the images. Similarity search for images is performed by determining the vectors with the smallest distances to a query vector. However, high retrieval quality does not depend on the actual distances of the feature vectors, but rather on the ranking order of the feature vectors from similar images. In this paper, we introduce a metric learning algorithm that focuses on identifying and modifying those feature vectors that most strongly affect the retrieval quality. We compute normalized approximated ranks and convert them to similarities by applying a nonlinear transfer function. These similarities are used in a newly proposed loss function that better contracts similar and disperses dissimilar samples. Experiments demonstrate significant improvement over existing deep feature embedding methods on the CUB-200-2011, Cars196, and Stanford Online Products data sets for all embedding sizes.

up
0 users have voted:

Paper Details

Authors:
Konstantin Schall, Kai Uwe Barthel, Nico Hezel, and Klaus Jung
Submitted On:
27 September 2019 - 10:06pm
Short Link:
Type:
Presentation Slides
Event:
Presenter's Name:
Kai Uwe Barthel
Document Year:
2019
Cite

Document Files

Nonlinear Rank Approximation Loss

(11)

Subscribe

[1] Konstantin Schall, Kai Uwe Barthel, Nico Hezel, and Klaus Jung, "Deep Metric Learning using Similarities from 
 Nonlinear Rank Approximations", IEEE SigPort, 2019. [Online]. Available: http://sigport.org/4847. Accessed: Oct. 18, 2019.
@article{4847-19,
url = {http://sigport.org/4847},
author = {Konstantin Schall; Kai Uwe Barthel; Nico Hezel; and Klaus Jung },
publisher = {IEEE SigPort},
title = {Deep Metric Learning using Similarities from 
 Nonlinear Rank Approximations},
year = {2019} }
TY - EJOUR
T1 - Deep Metric Learning using Similarities from 
 Nonlinear Rank Approximations
AU - Konstantin Schall; Kai Uwe Barthel; Nico Hezel; and Klaus Jung
PY - 2019
PB - IEEE SigPort
UR - http://sigport.org/4847
ER -
Konstantin Schall, Kai Uwe Barthel, Nico Hezel, and Klaus Jung. (2019). Deep Metric Learning using Similarities from 
 Nonlinear Rank Approximations. IEEE SigPort. http://sigport.org/4847
Konstantin Schall, Kai Uwe Barthel, Nico Hezel, and Klaus Jung, 2019. Deep Metric Learning using Similarities from 
 Nonlinear Rank Approximations. Available at: http://sigport.org/4847.
Konstantin Schall, Kai Uwe Barthel, Nico Hezel, and Klaus Jung. (2019). "Deep Metric Learning using Similarities from 
 Nonlinear Rank Approximations." Web.
1. Konstantin Schall, Kai Uwe Barthel, Nico Hezel, and Klaus Jung. Deep Metric Learning using Similarities from 
 Nonlinear Rank Approximations [Internet]. IEEE SigPort; 2019. Available from : http://sigport.org/4847