Scaling Limit of Neural Networks with the Xavier Initialization and Convergence to a Global Minimum.
Justin A. SirignanoKonstantinos SpiliopoulosPublished in: CoRR (2019)
Keyphrases
- global minimum
- global minima
- neural network
- cost function
- energy function
- energy minimization
- simulated annealing
- initial conditions
- weight update
- global optimum
- pattern recognition
- global minimization
- energy functional
- genetic algorithm
- error function
- initial guess
- artificial neural networks
- back propagation
- graph cuts
- feedforward neural networks
- global optimization
- convergence speed
- optimization algorithm
- markov random field
- global optimality
- prior knowledge
- image segmentation
- global minimizers