Matrix and tensor decompositions for training binary neural networks.
Adrian BulatJean KossaifiGeorgios TzimiropoulosMaja PanticPublished in: CoRR (2019)
Keyphrases
- neural network
- training process
- training algorithm
- singular value decomposition
- feed forward neural networks
- singular vectors
- feedforward neural networks
- multi layer perceptron
- pattern recognition
- back propagation
- trace norm
- structure tensor
- backpropagation algorithm
- neural network structure
- neural network training
- artificial neural networks
- genetic algorithm
- tensor factorization
- projection matrices
- dimensionality reduction
- tensor decomposition
- recurrent networks
- multilayer neural network
- higher order
- singular values
- linear algebra
- activation function
- anisotropic diffusion
- multilayer perceptron
- covariance matrix
- training set
- symmetric positive definite
- training data