DCBT-Net: Training Deep Convolutional Neural Networks With Extremely Noisy Labels.
Bekhzod OlimovJeonghong KimAnand PaulPublished in: IEEE Access (2020)
Keyphrases
- convolutional neural networks
- training set
- training examples
- convolutional network
- labelled data
- training data
- annotation effort
- deep architectures
- learning algorithm
- test set
- class labels
- training process
- training algorithm
- missing data
- fully labeled
- label propagation
- noisy data
- data sets
- supervised learning
- object detection
- multiscale
- decision trees
- neural network