Invariance-inducing regularization using worst-case transformations suffices to boost accuracy and spatial robustness.
Fanny YangZuowen WangChristina Heinze-DemlPublished in: CoRR (2019)
Keyphrases
- worst case
- high accuracy
- computational efficiency
- image transformations
- np hard
- upper bound
- computational speed
- classification accuracy
- space time
- spatial and temporal
- geometric and photometric transformations
- average case
- spatial data
- error rate
- lower bound
- error bounds
- spatial databases
- highly accurate
- spatial distribution
- computational cost
- invariance properties
- training data