Mining on Manifolds: Metric Learning without Labels
The result's identifiers
Result code in IS VaVaI
<a href="https://www.isvavai.cz/riv?ss=detail&h=RIV%2F68407700%3A21230%2F18%3A00327169" target="_blank" >RIV/68407700:21230/18:00327169 - isvavai.cz</a>
Result on the web
<a href="https://ieeexplore.ieee.org/document/8578895" target="_blank" >https://ieeexplore.ieee.org/document/8578895</a>
DOI - Digital Object Identifier
<a href="http://dx.doi.org/10.1109/CVPR.2018.00797" target="_blank" >10.1109/CVPR.2018.00797</a>
Alternative languages
Result language
angličtina
Original language name
Mining on Manifolds: Metric Learning without Labels
Original language description
In this work we present a novel unsupervised framework for hard training example mining. The only input to the method is a collection of images relevant to the target application and a meaningful initial representation, provided e.g. by pre-trained CNN. Positive examples are distant points on a single manifold, while negative examples are nearby points on different manifolds. Both types of examples are revealed by disagreements between Euclidean and manifold similarities. The discovered examples can be used in training with any discriminative loss. The method is applied to unsupervised fine-tuning of pre-trained networks for fine-grained classification and particular object retrieval. Our models are on par or are outperforming prior models that are fully or partially supervised.
Czech name
—
Czech description
—
Classification
Type
D - Article in proceedings
CEP classification
—
OECD FORD branch
10201 - Computer sciences, information science, bioinformathics (hardware development to be 2.2, social aspect to be 5.8)
Result continuities
Project
<a href="/en/project/LL1303" target="_blank" >LL1303: Large Scale Category Retrieval</a><br>
Continuities
P - Projekt vyzkumu a vyvoje financovany z verejnych zdroju (s odkazem do CEP)
Others
Publication year
2018
Confidentiality
S - Úplné a pravdivé údaje o projektu nepodléhají ochraně podle zvláštních právních předpisů
Data specific for result type
Article name in the collection
CVPR 2018: Proceedings of the 2018 IEEE Conference on Computer Vision and Pattern Recognition
ISBN
978-1-5386-6420-9
ISSN
1063-6919
e-ISSN
2575-7075
Number of pages
10
Pages from-to
7642-7651
Publisher name
IEEE
Place of publication
Piscataway, NJ
Event location
Salt Lake City
Event date
Jun 19, 2018
Type of event by nationality
WRD - Celosvětová akce
UT code for WoS article
000457843607082