Provable Identifiability of Two-Layer ReLU Neural Networks via LASSO Regularization.
Gen LiGanghua WangJie DingPublished in: CoRR (2023)
Keyphrases
- neural network
- multi layer
- regularization parameter
- single layer
- elastic net
- group lasso
- pattern recognition
- artificial neural networks
- genetic algorithm
- neural nets
- back propagation
- feature selection
- feed forward
- cross validation
- prior information
- recurrent neural networks
- sparse representation
- model selection
- sample size
- neural network model
- fault diagnosis
- fuzzy systems
- image restoration
- activation function
- multiple kernel learning
- self organizing maps
- causal models
- restricted boltzmann machine
- fuzzy logic