AugRmixAT: A Data Processing and Training Method for Improving Multiple Robustness and Generalization Performance.
Xiaoliang LiuFurao ShenJian ZhaoChanghai NiePublished in: ICME (2022)
Keyphrases
- data processing
- high accuracy
- synthetic data
- preprocessing
- experimental evaluation
- computational cost
- cost function
- significant improvement
- high precision
- training phase
- databases
- data analysis
- similarity measure
- optimization algorithm
- classification method
- error rate
- classification accuracy
- computational efficiency
- clustering method
- prior knowledge
- pairwise
- image segmentation