L-RED: Efficient Post-Training Detection of Imperceptible Backdoor Attacks without Access to the Training Set.
Zhen XiangDavid J. MillerGeorge KesidisPublished in: CoRR (2020)
Keyphrases
- training set
- test set
- watermarking scheme
- training examples
- training samples
- classification accuracy
- decision trees
- test data
- supervised learning
- nearest neighbor
- active learning
- feature space
- training data
- object detection
- anomaly detection
- detection algorithm
- false alarms
- data sets
- attack detection
- digital images
- false positives
- cross validation
- detection rate
- watermarking algorithm