INN: A Method Identifying Clean-annotated Samples via Consistency Effect in Deep Neural Networks.
Dongha KimYongchan ChoiKunwoong KimYongdai KimPublished in: CoRR (2021)
Keyphrases
- neural network
- high precision
- detection method
- high accuracy
- computational cost
- significant improvement
- support vector machine
- synthetic data
- dynamic programming
- experimental evaluation
- prior knowledge
- pairwise
- training samples
- preprocessing
- computational complexity
- neural network model
- classification method
- objective function
- fully automatic
- auto associative
- data sets
- neuro fuzzy systems
- bp neural network
- energy function
- detection algorithm
- optimization algorithm
- edge detection