Assessing Neural Network Representations During Training Using Noise-Resilient Diffusion Spectral Entropy.
Danqi LiaoChen LiuBenjamin W. ChristensenAlexander TongGuillaume HuguetGuy WolfMaximilian NickelIan AdelsteinSmita KrishnaswamyPublished in: CoRR (2023)
Keyphrases
- neural network
- train a neural network
- training algorithm
- training process
- feed forward neural networks
- multi layer perceptron
- feedforward neural networks
- artificial neural networks
- neural network training
- trained neural network
- noise level
- information theory
- back propagation
- spectral analysis
- training patterns
- backpropagation algorithm
- neural network structure
- genetic algorithm
- support vector machine
- mutual information
- noisy data
- image segmentation
- training and testing data
- training set
- training samples
- missing data
- noise model
- feed forward
- gaussian noise
- training phase
- median filter
- anisotropic diffusion
- activation function
- bp neural network
- radial basis function
- information theoretic
- fuzzy neural network
- noise reduction
- hidden layer
- neural network is trained
- spectral resolution
- recurrent neural networks
- multilayer neural network