on the defense against adversarial examples beyond the
play

On the Defense Against Adversarial Examples Beyond the Visible - PowerPoint PPT Presentation

On the Defense Against Adversarial Examples Beyond the Visible Spectrum Anthony Ortiz 1 , Olac Fuentes 1 , Dalton Rosario 2 , Christopher Kiekintveld 1 1 Department of Computer Science, UTEP 2 US Army Research Laboratory Adversarial Examples on


  1. On the Defense Against Adversarial Examples Beyond the Visible Spectrum Anthony Ortiz 1 , Olac Fuentes 1 , Dalton Rosario 2 , Christopher Kiekintveld 1 1 Department of Computer Science, UTEP 2 US Army Research Laboratory

  2. Adversarial Examples on Natural Images 2

  3. Adversarial Examples Beyond the Visible Spectrum 3

  4. Experimantal Setup •DSTL Dataset: •1 km x 1 km Satellite Image •Spatial resolution: 31 cm •3 channels RGB •8 Channels VNIR •8 Channels SWIR •10 Classes (Buildings, roads, track, trees, crops) •DigitalGlobe’s WorldView Satellite System •Task: Semantic Segmentation •Evaluation Metric: Mean IoU •Architecture: •Fully Convolutional Networks (FCN-8) with VGG-19 as backbone 4

  5. Performance Evaluation DSTL Dataset 5

  6. Adversarial Examples Beyond Visible Spectrum 6

  7. Adversarial Examples Beyond Visible Spectrum 7

  8. Adversarial Examples Beyond Visible Spectrum 8

  9. Dynamic Adversarial Perturbation Attack True Color Input Prediction Clean Prediction Adversarial 9

  10. ILFS as a Defense Against Adversarial Examples 10

  11. Detecting Adversarial Examples 11

  12. Spectral Signature Adversarial Examples FGSM Iterative FGSM 12

  13. Wetness Index Band swir2:1550-1590nm Band swir4: 1710-1750nm 13

  14. Detector Network Architecture 14

  15. Detection Results 15

  16. Adversarial Training Helps 16

  17. Adversarial Training Helps 17

  18. Conclusions ● Multispectral and Hyperspectral Images are vulnerable to adversarial examples. With the right prior, adversarial examples can successfully be detected. ● ● Adversarial Training improve models robustness beyond RGB and generalize across attacks. 18

  19. Thank you 19

Recommend


More recommend