An Empirical Study of the Hidden Matrix Rank for Neural Networks with Random Weights.
Pablo A. HenríquezGonzalo A. RuzPublished in: ICMLA (2017)
Keyphrases
- neural network
- singular values
- pattern recognition
- nuclear norm minimization
- hopfield neural network
- back propagation
- weight update
- weight matrix
- artificial neural networks
- linear combination
- singular value decomposition
- weighted sum
- genetic algorithm
- moore penrose
- multi layer
- multilayer perceptron
- neural nets
- fuzzy logic
- hidden neurons
- feed forward
- connection weights
- rank order
- recurrent neural networks
- weighting scheme
- activation function
- ranked list
- network architecture
- low rank
- relative importance
- matrix representation
- training process
- fuzzy systems
- neural network model
- objective function