Neural networks trained with SGD learn distributions of increasing complexity.
Maria RefinettiAlessandro IngrossoSebastian GoldtPublished in: ICML (2023)
Keyphrases
- neural network
- multilayer perceptron
- pattern recognition
- trained neural networks
- trained neural network
- multi layer perceptron
- training process
- artificial neural networks
- learning rules
- space complexity
- fuzzy logic
- neural network model
- probability distribution
- neural nets
- rule extraction
- auto associative
- computational complexity
- genetic algorithm
- self organizing maps
- feed forward
- multi layer
- training set
- heavy tailed