Multimodal similarity-preserving hashing

Jonathan Masci; Michael M. Bronstein; Alexander M. Bronstein; Jurgen Schmidhuber

doi:10.1109/TPAMI.2013.225

Multimodal similarity-preserving hashing

Jonathan Masci, Michael M. Bronstein, Alexander M. Bronstein, Jurgen Schmidhuber

Research output: Contribution to journal › Article › peer-review

Abstract

We introduce an efficient computational framework for hashing data belonging to multiple modalities into a single representation space where they become mutually comparable. The proposed approach is based on a novel coupled siamese neural network architecture and allows unified treatment of intra-and inter-modality similarity learning. Unlike existing cross-modality similarity learning approaches, our hashing functions are not limited to binarized linear projections and can assume arbitrarily complex forms. We show experimentally that our method significantly outperforms state-of-the-art hashing approaches on multimedia retrieval tasks.

Original language	English
Article number	6654144
Pages (from-to)	824-830
Number of pages	7
Journal	IEEE Transactions on Pattern Analysis and Machine Intelligence
Volume	36
Issue number	4
DOIs	https://doi.org/10.1109/TPAMI.2013.225
State	Published - Apr 2014
Externally published	Yes

Keywords

Similarity-sensitive hashing
feature descriptor
metric learning
neural network

All Science Journal Classification (ASJC) codes

Software
Computer Vision and Pattern Recognition
Computational Theory and Mathematics
Artificial Intelligence
Applied Mathematics

Access to Document

10.1109/TPAMI.2013.225

Cite this

@article{0602aedbc3c5459f9c623d1a7fb55953,

title = "Multimodal similarity-preserving hashing",

abstract = "We introduce an efficient computational framework for hashing data belonging to multiple modalities into a single representation space where they become mutually comparable. The proposed approach is based on a novel coupled siamese neural network architecture and allows unified treatment of intra-and inter-modality similarity learning. Unlike existing cross-modality similarity learning approaches, our hashing functions are not limited to binarized linear projections and can assume arbitrarily complex forms. We show experimentally that our method significantly outperforms state-of-the-art hashing approaches on multimedia retrieval tasks.",

keywords = "Similarity-sensitive hashing, feature descriptor, metric learning, neural network",

author = "Jonathan Masci and Bronstein, \{Michael M.\} and Bronstein, \{Alexander M.\} and Jurgen Schmidhuber",

year = "2014",

month = apr,

doi = "10.1109/TPAMI.2013.225",

language = "الإنجليزيّة",

volume = "36",

pages = "824--830",

journal = "IEEE Transactions on Pattern Analysis and Machine Intelligence",

issn = "0162-8828",

publisher = "IEEE Computer Society",

number = "4",

}

TY - JOUR

T1 - Multimodal similarity-preserving hashing

AU - Masci, Jonathan

AU - Bronstein, Michael M.

AU - Bronstein, Alexander M.

AU - Schmidhuber, Jurgen

PY - 2014/4

Y1 - 2014/4

N2 - We introduce an efficient computational framework for hashing data belonging to multiple modalities into a single representation space where they become mutually comparable. The proposed approach is based on a novel coupled siamese neural network architecture and allows unified treatment of intra-and inter-modality similarity learning. Unlike existing cross-modality similarity learning approaches, our hashing functions are not limited to binarized linear projections and can assume arbitrarily complex forms. We show experimentally that our method significantly outperforms state-of-the-art hashing approaches on multimedia retrieval tasks.

AB - We introduce an efficient computational framework for hashing data belonging to multiple modalities into a single representation space where they become mutually comparable. The proposed approach is based on a novel coupled siamese neural network architecture and allows unified treatment of intra-and inter-modality similarity learning. Unlike existing cross-modality similarity learning approaches, our hashing functions are not limited to binarized linear projections and can assume arbitrarily complex forms. We show experimentally that our method significantly outperforms state-of-the-art hashing approaches on multimedia retrieval tasks.

KW - Similarity-sensitive hashing

KW - feature descriptor

KW - metric learning

KW - neural network

UR - http://www.scopus.com/inward/record.url?scp=84897493967&partnerID=8YFLogxK

U2 - 10.1109/TPAMI.2013.225

DO - 10.1109/TPAMI.2013.225

M3 - مقالة

SN - 0162-8828

VL - 36

SP - 824

EP - 830

JO - IEEE Transactions on Pattern Analysis and Machine Intelligence

JF - IEEE Transactions on Pattern Analysis and Machine Intelligence

IS - 4

M1 - 6654144

ER -

Multimodal similarity-preserving hashing

Abstract

Keywords

All Science Journal Classification (ASJC) codes

Access to Document

Other files and links

Fingerprint

Cite this