Deep networks with ReLU activation functions can be smooth statistical models.
Joseph RynkiewiczPublished in: ESANN (2022)
Keyphrases
- statistical models
- activation function
- network size
- statistical model
- statistical modeling
- neural network
- feed forward
- hidden layer
- neural architecture
- parameter estimation
- artificial neural networks
- neural nets
- back propagation
- multilayer perceptron
- network architecture
- radial basis function
- network structure
- social networks
- learning rate
- basis functions
- genetic algorithm