Adapting Newton's Method to Neural Networks through a Summary of Higher-Order Derivatives.
Pierre WolinskiPublished in: CoRR (2023)
Keyphrases
- higher order
- neural network
- pairwise
- cost function
- high accuracy
- directional derivatives
- rule extraction
- high precision
- segmentation method
- synthetic data
- fuzzy logic
- computational cost
- significant improvement
- objective function
- feature extraction
- markov random field
- input data
- detection method
- preprocessing
- fault diagnosis
- high order
- neural nets
- image segmentation