On the training of sparse and dense deep neural networks: less parameters, same performance.
Lorenzo ChicchiLorenzo GiambagliLorenzo BuffoniTimoteo CarlettiMarco CiavarellaDuccio FanelliPublished in: CoRR (2021)
Keyphrases
- neural network
- training process
- radial basis function network
- training algorithm
- training set
- parameter values
- pattern recognition
- data sets
- neural network training
- feedforward neural networks
- feed forward neural networks
- parameter space
- self organizing maps
- back propagation
- training examples
- artificial neural networks
- fault diagnosis
- dense stereo
- input data
- fuzzy logic
- initial estimates
- deep architectures
- high dimensional
- genetic algorithm
- motion field estimation
- test set
- neural network structure
- face recognition
- sparse data
- multi layer perceptron
- training phase
- parameter settings
- linear combination