AugRmixAT: A Data Processing and Training Method for Improving Multiple Robustness and Generalization Performance.
Xiaoliang LiuFurao ShenJian ZhaoChanghai NiePublished in: CoRR (2022)
Keyphrases
- data processing
- computational efficiency
- training phase
- training samples
- high accuracy
- experimental evaluation
- high precision
- computational cost
- synthetic data
- detection method
- training process
- preprocessing
- error rate
- support vector machine
- classification method
- segmentation method
- optimization algorithm
- support vector machine svm
- training algorithm
- edge detection
- image quality
- databases
- supervised learning
- classification accuracy
- probabilistic model
- dynamic programming
- cost function
- significant improvement
- prior knowledge
- objective function
- similarity measure
- machine learning