Circular Shift: An Effective Data Augmentation Method For Convolutional Neural Network On Image Classification.
Kailai ZhangZheng CaoJi WuPublished in: ICIP (2020)
Keyphrases
- synthetic data
- image classification
- noisy data
- high quality
- test data
- missing data
- prior knowledge
- input data
- database
- information loss
- preprocessing
- convolutional neural network
- statistical methods
- data collection
- data sets
- data sources
- similarity measure
- data mining techniques
- data quality
- original data
- missing values
- detection method
- training samples
- high dimensional
- object recognition
- data structure
- detection algorithm
- objective function
- prior information
- spectral clustering
- significant improvement
- probabilistic model
- image processing
- correlation analysis
- image features