Why adversarial training can hurt robust accuracy.
Jacob ClarysseJulia HörmannFanny YangPublished in: CoRR (2022)
Keyphrases
- high accuracy
- training set
- computational cost
- highly accurate
- supervised learning
- classification accuracy
- error rate
- multi agent
- prediction accuracy
- test set
- training process
- high precision
- high computational efficiency
- classification performances
- training set size
- parameter tuning
- partial occlusion
- precision and recall
- computational efficiency
- computational complexity