Bayes-SAR net: Robust SAR image classification with uncertainty estimation using Bayesian convolutional neural network

Dimah Dera, Ghulam Rasool, Nidhal C. Bouaynaya, Adam Eichen, Stephen Shanko, Jeff Cammerata, Sanipa Arnold

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

Abstract

Synthetic aperture radar (SAR) image classification is a challenging problem due to the complex imaging mechanism as well as the random speckle noise, which affects radar image interpretation. Recently, convolutional neural networks (CNNs) have been shown to outperform previous state-of-the-art techniques in computer vision tasks owing to their ability to learn relevant features from the data. However, CNNs in particular and neural networks, in general, lack uncertainty quantification and can be easily deceived by adversarial attacks. This paper proposes Bayes-SAR Net, a Bayesian CNN that can perform robust SAR image classification while quantifying the uncertainty or confidence of the network in its decision. Bayes-SAR Net propagates the first two moments (mean and covariance) of the approximate posterior distribution of the network parameters given the data and obtains a predictive mean and covariance of the classification output. Experiments, using the benchmark datasets Flevoland and Oberpfaffenhofen, show superior performance and robustness to Gaussian noise and adversarial attacks, as compared to the SAR-Net homologue. Bayes-SAR Net achieves a test accuracy that is around 10% higher in the case of adversarial perturbation (levels ≽ 0.05).

Original languageEnglish (US)
Title of host publication2020 IEEE International Radar Conference, RADAR 2020
PublisherInstitute of Electrical and Electronics Engineers Inc.
Pages362-367
Number of pages6
ISBN (Electronic)9781728168128
DOIs
StatePublished - Apr 2020
Event2020 IEEE International Radar Conference, RADAR 2020 - Washington, United States
Duration: Apr 28 2020Apr 30 2020

Publication series

Name2020 IEEE International Radar Conference, RADAR 2020

Conference

Conference2020 IEEE International Radar Conference, RADAR 2020
CountryUnited States
CityWashington
Period4/28/204/30/20

All Science Journal Classification (ASJC) codes

  • Computer Networks and Communications
  • Signal Processing
  • Instrumentation

Fingerprint Dive into the research topics of 'Bayes-SAR net: Robust SAR image classification with uncertainty estimation using Bayesian convolutional neural network'. Together they form a unique fingerprint.

Cite this