Completed
Introduction
Class Central Classrooms beta
YouTube playlists curated by Class Central.
Classroom Contents
Trust Region Based Adversarial Attack on Neural Networks
Automatically move to the next video in the Classroom when playback concludes
- 1 Introduction
- 2 3D adversarial objects
- 3 Physical attacks on traffic signs
- 4 Adversarial Patches to Attack Person Detection
- 5 Semantic Segmentation and Object Detection
- 6 LIDAR attack
- 7 Definitions
- 8 Problem Formulation
- 9 Fast Gradient Sign Method • Goodfellow et al. proposed the Fast Gradient Sign Method FGSM
- 10 Basic Iterative Method
- 11 Carlini-Wagner Attack
- 12 Problems and Challenges
- 13 Other Problems
- 14 Contributions
- 15 Trust Region Optimization
- 16 Updating the Trust Region
- 17 Trust region example - Initial Start
- 18 Iteration 1
- 19 Final Trajectory after 20 iterations
- 20 Proposed Method
- 21 Metrics
- 22 Types of attacks used
- 23 Summary of setup
- 24 Time Performance on ImageNet
- 25 Qualitative Results
- 26 ImageNet Results
- 27 Second order attack results
- 28 Conclusion
