PULP-NN: Accelerating Quantized Neural Networks on Parallel Ultra-Low-Power RISC-V Processors.
Angelo GarofaloManuele RusciFrancesco ContiDavide RossiLuca BeniniPublished in: CoRR (2019)
Keyphrases
- neural network
- parallel processing
- instruction set
- shared memory
- single processor
- parallel computation
- parallel computing
- parallel algorithm
- artificial neural networks
- distributed memory
- parallel execution
- ultra low power
- parallel programming
- multiprocessor systems
- computer architecture
- back propagation
- pattern recognition
- data parallelism
- neural network model
- processing elements
- application specific
- multilayer perceptron
- level parallelism
- parallel architectures
- multi layer
- multi core processors
- feedforward neural networks
- feed forward
- parallel version
- parallel architecture
- hardware architecture
- multithreading
- genetic algorithm
- knn
- fault diagnosis
- feed forward neural networks
- modular neural networks
- processing units
- fuzzy logic
- nearest neighbor
- message passing interface
- parallel computers
- parallel processors
- low power
- self organizing maps
- radial basis function
- bp neural network
- low power consumption
- general purpose
- hardware implementation
- high end
- recurrent neural networks
- associative memory
- high performance computing
- high speed
- training process
- hidden layer
- computing systems
- weight update
- floating point
- massively parallel
- graphics processing units