Deep linear neural networks with arbitrary loss: All local minima are global.
Thomas LaurentJames von BrechtPublished in: CoRR (2017)
Keyphrases
- real time
- neural network
- back propagation
- genetic algorithm
- highly non linear
- data sets
- fault diagnosis
- user defined
- neural network model
- recurrent neural networks
- neural nets
- artificial neural networks
- simulated annealing
- evolutionary algorithm
- self organizing maps
- closed form
- multi layer
- fuzzy systems
- network architecture
- linear systems
- linear constraints
- deep learning