Typical and atypical solutions in non-convex neural networks with discrete and continuous weights.
Carlo BaldassiEnrico M. MalatestaGabriele PeruginiRiccardo ZecchinaPublished in: CoRR (2023)
Keyphrases
- neural network
- discrete space
- piecewise linear
- continuous functions
- global exponential stability
- continuous space
- continuous variables
- continuous action
- artificial neural networks
- fuzzy logic
- discrete geometry
- continuous domains
- back propagation
- convex hull
- pattern recognition
- real world
- discrete data
- weighted sum
- optimal solution
- self organizing maps
- benchmark problems
- hopfield neural network
- genetic algorithm
- multilayer perceptron
- continuous data
- convex optimization
- convex sets
- differential equations
- neural nets
- network architecture
- recurrent neural networks
- linear program
- discrete variables
- relative importance
- hidden neurons
- finite number