Persistent RNNs: Stashing Recurrent Weights On-Chip.
Greg DiamosShubho SenguptaBryan CatanzaroMike ChrzanowskiAdam CoatesErich ElsenJesse H. EngelAwni Y. HannunSanjeev SatheeshPublished in: ICML (2016)
Keyphrases
- recurrent neural networks
- feed forward
- feedforward neural networks
- neural network
- high speed
- recurrent networks
- low cost
- linear combination
- weighted sum
- neuro fuzzy
- hidden layer
- artificial neural networks
- analog vlsi
- high density
- relative importance
- echo state networks
- weighting scheme
- physical design
- vlsi implementation
- database
- database systems
- fuzzy neural network
- optimal weights
- single chip
- information retrieval
- genetic algorithm
- circuit design
- infrared
- rough sets