AutoAttacker: A reinforcement learning approach for black-box adversarial attacks.
Ilias TsingenopoulosDavy PreuveneersWouter JoosenPublished in: EuroS&P Workshops (2019)
Keyphrases
- black box
- reinforcement learning
- multi agent
- black boxes
- function approximation
- countermeasures
- white box
- integration testing
- hybrid systems
- state space
- watermarking scheme
- learning algorithm
- model free
- rule extraction
- test cases
- state transition
- temporal difference
- reinforcement learning algorithms
- neural network
- white box testing
- action space
- learning capabilities
- high level
- multimedia
- machine learning