How to Train Your Wide Neural Network Without Backprop: An Input-Weight Alignment Perspective.
Akhilan BoopathyIla FietePublished in: ICML (2022)
Keyphrases
- neural network
- desired output
- pattern recognition
- weight update
- input data
- multi layer
- fuzzy logic
- image alignment
- self organizing maps
- neural network model
- neural network is trained
- learning vector quantization
- bp neural network
- recurrent neural networks
- wide range
- back propagation
- sequence alignment
- artificial neural networks
- word alignment
- expert systems
- genetic algorithm