Neural Network Compression using Binarization and Few Full-Precision Weights.
Franco Maria NardiniCosimo RulliSalvatore TraniRossano VenturiniPublished in: CoRR (2023)
Keyphrases
- neural network
- kohonen self organizing map
- connection weights
- image compression
- document images
- hidden neurons
- weighted sum
- weight update
- high precision
- compression algorithm
- network architecture
- pattern recognition
- back propagation
- relative importance
- genetic algorithm
- bp neural network
- data compression
- artificial neural networks
- fuzzy logic
- linear combination
- precision and recall
- neural network is trained
- neural network model
- neural model
- feed forward neural networks
- neural nets
- compression scheme
- weighting scheme
- activation function
- average precision
- multilayer perceptron
- feedforward neural networks
- grayscale images
- gray level
- image preprocessing
- multi layer perceptron
- associative memory