Passive millimeter wave image classification with large scale Gaussian processes

Pablo Morales, Adrian Perez-Suay, Rafael Molina, Gustau Camps-Valls, Aggelos K Katsaggelos

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

Passive Millimeter Wave Images (PMMWIs) are being increasingly used to identify and localize objects concealed under clothing. Taking into account the quality of these images and the unknown position, shape, and size of the hidden objects, large data sets are required to build successful classification/detection systems. Kernel methods, in particular Gaussian Processes (GPs), are sound, flexible, and popular techniques to address supervised learning problems. Unfortunately, their computational cost is known to be prohibitive for large scale applications. In this work, we present a novel approach to PMMWI classification based on the use of Gaussian Processes for large data sets. The proposed methodology relies on linear approximations to kernel functions through random Fourier features. Model hyperparameters are learned within a variational Bayes inference scheme. Our proposal is well suited for real-time applications, since its computational cost at training and test times is much lower than the original GP formulation. The proposed approach is tested on a unique, large, and real PMMWI database containing a broad variety of sizes, types, and locations of hidden objects.

Original languageEnglish (US)
Title of host publication2017 IEEE International Conference on Image Processing, ICIP 2017 - Proceedings
PublisherIEEE Computer Society
Pages370-374
Number of pages5
Volume2017-September
ISBN (Electronic)9781509021758
DOIs
StatePublished - Feb 20 2018
Event24th IEEE International Conference on Image Processing, ICIP 2017 - Beijing, China
Duration: Sep 17 2017Sep 20 2017

Other

Other24th IEEE International Conference on Image Processing, ICIP 2017
CountryChina
CityBeijing
Period9/17/179/20/17

Fingerprint

Image classification
Millimeter waves
Supervised learning
Costs
Acoustic waves

Keywords

  • Gaussian processes
  • Large scale classification
  • PMMWI
  • Random Fourier features
  • Variational inference

ASJC Scopus subject areas

  • Software
  • Computer Vision and Pattern Recognition
  • Signal Processing

Cite this

Morales, P., Perez-Suay, A., Molina, R., Camps-Valls, G., & Katsaggelos, A. K. (2018). Passive millimeter wave image classification with large scale Gaussian processes. In 2017 IEEE International Conference on Image Processing, ICIP 2017 - Proceedings (Vol. 2017-September, pp. 370-374). IEEE Computer Society. https://doi.org/10.1109/ICIP.2017.8296305
Morales, Pablo ; Perez-Suay, Adrian ; Molina, Rafael ; Camps-Valls, Gustau ; Katsaggelos, Aggelos K. / Passive millimeter wave image classification with large scale Gaussian processes. 2017 IEEE International Conference on Image Processing, ICIP 2017 - Proceedings. Vol. 2017-September IEEE Computer Society, 2018. pp. 370-374
@inproceedings{1266fa24668f418db5501034026bd130,
title = "Passive millimeter wave image classification with large scale Gaussian processes",
abstract = "Passive Millimeter Wave Images (PMMWIs) are being increasingly used to identify and localize objects concealed under clothing. Taking into account the quality of these images and the unknown position, shape, and size of the hidden objects, large data sets are required to build successful classification/detection systems. Kernel methods, in particular Gaussian Processes (GPs), are sound, flexible, and popular techniques to address supervised learning problems. Unfortunately, their computational cost is known to be prohibitive for large scale applications. In this work, we present a novel approach to PMMWI classification based on the use of Gaussian Processes for large data sets. The proposed methodology relies on linear approximations to kernel functions through random Fourier features. Model hyperparameters are learned within a variational Bayes inference scheme. Our proposal is well suited for real-time applications, since its computational cost at training and test times is much lower than the original GP formulation. The proposed approach is tested on a unique, large, and real PMMWI database containing a broad variety of sizes, types, and locations of hidden objects.",
keywords = "Gaussian processes, Large scale classification, PMMWI, Random Fourier features, Variational inference",
author = "Pablo Morales and Adrian Perez-Suay and Rafael Molina and Gustau Camps-Valls and Katsaggelos, {Aggelos K}",
year = "2018",
month = "2",
day = "20",
doi = "10.1109/ICIP.2017.8296305",
language = "English (US)",
volume = "2017-September",
pages = "370--374",
booktitle = "2017 IEEE International Conference on Image Processing, ICIP 2017 - Proceedings",
publisher = "IEEE Computer Society",
address = "United States",

}

Morales, P, Perez-Suay, A, Molina, R, Camps-Valls, G & Katsaggelos, AK 2018, Passive millimeter wave image classification with large scale Gaussian processes. in 2017 IEEE International Conference on Image Processing, ICIP 2017 - Proceedings. vol. 2017-September, IEEE Computer Society, pp. 370-374, 24th IEEE International Conference on Image Processing, ICIP 2017, Beijing, China, 9/17/17. https://doi.org/10.1109/ICIP.2017.8296305

Passive millimeter wave image classification with large scale Gaussian processes. / Morales, Pablo; Perez-Suay, Adrian; Molina, Rafael; Camps-Valls, Gustau; Katsaggelos, Aggelos K.

2017 IEEE International Conference on Image Processing, ICIP 2017 - Proceedings. Vol. 2017-September IEEE Computer Society, 2018. p. 370-374.

Research output: Chapter in Book/Report/Conference proceedingConference contribution

TY - GEN

T1 - Passive millimeter wave image classification with large scale Gaussian processes

AU - Morales, Pablo

AU - Perez-Suay, Adrian

AU - Molina, Rafael

AU - Camps-Valls, Gustau

AU - Katsaggelos, Aggelos K

PY - 2018/2/20

Y1 - 2018/2/20

N2 - Passive Millimeter Wave Images (PMMWIs) are being increasingly used to identify and localize objects concealed under clothing. Taking into account the quality of these images and the unknown position, shape, and size of the hidden objects, large data sets are required to build successful classification/detection systems. Kernel methods, in particular Gaussian Processes (GPs), are sound, flexible, and popular techniques to address supervised learning problems. Unfortunately, their computational cost is known to be prohibitive for large scale applications. In this work, we present a novel approach to PMMWI classification based on the use of Gaussian Processes for large data sets. The proposed methodology relies on linear approximations to kernel functions through random Fourier features. Model hyperparameters are learned within a variational Bayes inference scheme. Our proposal is well suited for real-time applications, since its computational cost at training and test times is much lower than the original GP formulation. The proposed approach is tested on a unique, large, and real PMMWI database containing a broad variety of sizes, types, and locations of hidden objects.

AB - Passive Millimeter Wave Images (PMMWIs) are being increasingly used to identify and localize objects concealed under clothing. Taking into account the quality of these images and the unknown position, shape, and size of the hidden objects, large data sets are required to build successful classification/detection systems. Kernel methods, in particular Gaussian Processes (GPs), are sound, flexible, and popular techniques to address supervised learning problems. Unfortunately, their computational cost is known to be prohibitive for large scale applications. In this work, we present a novel approach to PMMWI classification based on the use of Gaussian Processes for large data sets. The proposed methodology relies on linear approximations to kernel functions through random Fourier features. Model hyperparameters are learned within a variational Bayes inference scheme. Our proposal is well suited for real-time applications, since its computational cost at training and test times is much lower than the original GP formulation. The proposed approach is tested on a unique, large, and real PMMWI database containing a broad variety of sizes, types, and locations of hidden objects.

KW - Gaussian processes

KW - Large scale classification

KW - PMMWI

KW - Random Fourier features

KW - Variational inference

UR - http://www.scopus.com/inward/record.url?scp=85045320329&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85045320329&partnerID=8YFLogxK

U2 - 10.1109/ICIP.2017.8296305

DO - 10.1109/ICIP.2017.8296305

M3 - Conference contribution

VL - 2017-September

SP - 370

EP - 374

BT - 2017 IEEE International Conference on Image Processing, ICIP 2017 - Proceedings

PB - IEEE Computer Society

ER -

Morales P, Perez-Suay A, Molina R, Camps-Valls G, Katsaggelos AK. Passive millimeter wave image classification with large scale Gaussian processes. In 2017 IEEE International Conference on Image Processing, ICIP 2017 - Proceedings. Vol. 2017-September. IEEE Computer Society. 2018. p. 370-374 https://doi.org/10.1109/ICIP.2017.8296305