Emphasis Regularisation by Gradient Rescaling for Training Deep Neural Networks with Noisy Labels.
Xinshao WangYang HuaElyor KodirovNeil Martin RobertsonPublished in: CoRR (2019)
Keyphrases
- neural network
- training process
- training algorithm
- training set
- training examples
- feedforward neural networks
- feed forward neural networks
- backpropagation algorithm
- neural network training
- back propagation
- multi layer perceptron
- labelled data
- model selection
- training phase
- pattern recognition
- training data
- neural network structure
- feed forward
- deep architectures
- artificial neural networks
- error back propagation
- multilayer perceptron
- recurrent networks
- multilayer neural network
- annotation effort
- test set
- supervised learning
- fuzzy logic
- energy functional
- noisy data
- edge detection
- active learning
- label noise
- multiscale
- image segmentation
- weighted sums
- machine learning