Exponential Discriminative Metric Embedding in Deep Learning
With the remarkable success achieved by the Convolutional Neural Networks (CNNs) in object recognition recently, deep learning is being widely used in the computer vision community. Deep Metric Learning (DML), integrating deep learning with conventional metric learning, has set new records in many f...
Main Authors: | , , , |
---|---|
Format: | Report |
Language: | unknown |
Published: |
arXiv
2018
|
Subjects: | |
Online Access: | https://dx.doi.org/10.48550/arxiv.1803.02504 https://arxiv.org/abs/1803.02504 |
id |
ftdatacite:10.48550/arxiv.1803.02504 |
---|---|
record_format |
openpolar |
spelling |
ftdatacite:10.48550/arxiv.1803.02504 2023-05-15T16:01:20+02:00 Exponential Discriminative Metric Embedding in Deep Learning Wu, Bowen Chen, Zhangling Wang, Jun Wu, Huaming 2018 https://dx.doi.org/10.48550/arxiv.1803.02504 https://arxiv.org/abs/1803.02504 unknown arXiv arXiv.org perpetual, non-exclusive license http://arxiv.org/licenses/nonexclusive-distrib/1.0/ Computer Vision and Pattern Recognition cs.CV Machine Learning cs.LG Machine Learning stat.ML FOS Computer and information sciences Preprint Article article CreativeWork 2018 ftdatacite https://doi.org/10.48550/arxiv.1803.02504 2022-04-01T09:43:48Z With the remarkable success achieved by the Convolutional Neural Networks (CNNs) in object recognition recently, deep learning is being widely used in the computer vision community. Deep Metric Learning (DML), integrating deep learning with conventional metric learning, has set new records in many fields, especially in classification task. In this paper, we propose a replicable DML method, called Include and Exclude (IE) loss, to force the distance between a sample and its designated class center away from the mean distance of this sample to other class centers with a large margin in the exponential feature projection space. With the supervision of IE loss, we can train CNNs to enhance the intra-class compactness and inter-class separability, leading to great improvements on several public datasets ranging from object recognition to face verification. We conduct a comparative study of our algorithm with several typical DML methods on three kinds of networks with different capacity. Extensive experiments on three object recognition datasets and two face recognition datasets demonstrate that IE loss is always superior to other mainstream DML methods and approach the state-of-the-art results. Report DML DataCite Metadata Store (German National Library of Science and Technology) |
institution |
Open Polar |
collection |
DataCite Metadata Store (German National Library of Science and Technology) |
op_collection_id |
ftdatacite |
language |
unknown |
topic |
Computer Vision and Pattern Recognition cs.CV Machine Learning cs.LG Machine Learning stat.ML FOS Computer and information sciences |
spellingShingle |
Computer Vision and Pattern Recognition cs.CV Machine Learning cs.LG Machine Learning stat.ML FOS Computer and information sciences Wu, Bowen Chen, Zhangling Wang, Jun Wu, Huaming Exponential Discriminative Metric Embedding in Deep Learning |
topic_facet |
Computer Vision and Pattern Recognition cs.CV Machine Learning cs.LG Machine Learning stat.ML FOS Computer and information sciences |
description |
With the remarkable success achieved by the Convolutional Neural Networks (CNNs) in object recognition recently, deep learning is being widely used in the computer vision community. Deep Metric Learning (DML), integrating deep learning with conventional metric learning, has set new records in many fields, especially in classification task. In this paper, we propose a replicable DML method, called Include and Exclude (IE) loss, to force the distance between a sample and its designated class center away from the mean distance of this sample to other class centers with a large margin in the exponential feature projection space. With the supervision of IE loss, we can train CNNs to enhance the intra-class compactness and inter-class separability, leading to great improvements on several public datasets ranging from object recognition to face verification. We conduct a comparative study of our algorithm with several typical DML methods on three kinds of networks with different capacity. Extensive experiments on three object recognition datasets and two face recognition datasets demonstrate that IE loss is always superior to other mainstream DML methods and approach the state-of-the-art results. |
format |
Report |
author |
Wu, Bowen Chen, Zhangling Wang, Jun Wu, Huaming |
author_facet |
Wu, Bowen Chen, Zhangling Wang, Jun Wu, Huaming |
author_sort |
Wu, Bowen |
title |
Exponential Discriminative Metric Embedding in Deep Learning |
title_short |
Exponential Discriminative Metric Embedding in Deep Learning |
title_full |
Exponential Discriminative Metric Embedding in Deep Learning |
title_fullStr |
Exponential Discriminative Metric Embedding in Deep Learning |
title_full_unstemmed |
Exponential Discriminative Metric Embedding in Deep Learning |
title_sort |
exponential discriminative metric embedding in deep learning |
publisher |
arXiv |
publishDate |
2018 |
url |
https://dx.doi.org/10.48550/arxiv.1803.02504 https://arxiv.org/abs/1803.02504 |
genre |
DML |
genre_facet |
DML |
op_rights |
arXiv.org perpetual, non-exclusive license http://arxiv.org/licenses/nonexclusive-distrib/1.0/ |
op_doi |
https://doi.org/10.48550/arxiv.1803.02504 |
_version_ |
1766397238090661888 |