Conversion of artificial recurrent neural networks to spiking neural networks for low-power neuromorphic hardware.
Peter U. DiehlGuido ZarrellaAndrew CassidyBruno U. PedroniEmre NeftciPublished in: ICRC (2016)
Keyphrases
- low power
- recurrent neural networks
- spiking neural networks
- low cost
- feed forward
- single chip
- neural models
- vlsi architecture
- artificial neural networks
- digital signal processing
- neural network
- power consumption
- hebbian learning
- low power consumption
- high speed
- biologically plausible
- biologically inspired
- spiking neurons
- hardware and software
- image sensor
- hidden layer
- echo state networks
- neural model
- back propagation
- recurrent networks
- power reduction
- signal processor
- real time
- mixed signal
- learning rules
- embedded systems
- gate array
- visual cortex
- logic circuits
- receptive fields
- training algorithm
- cmos technology
- learning algorithm