Why Quantization Improves Generalization: NTK of Binary Weight Neural Networks.
Kaiqi ZhangMing YinYu-Xiang WangPublished in: CoRR (2022)
Keyphrases
- neural network
- pattern recognition
- weight update
- genetic algorithm
- code length
- artificial neural networks
- fuzzy logic
- multilayer perceptron
- machine learning
- gray code
- learning rules
- neural nets
- training process
- non binary
- rule extraction
- quantization error
- hidden nodes
- recurrent neural networks
- self organizing maps
- learning algorithm
- uniform quantization