A Programmable Neural-Network Inference Accelerator Based on Scalable In-Memory Computing.
Hongyang JiaMurat OzatayYinqi TangHossein ValaviRakshit PathakJinseok LeeNaveen VermaPublished in: ISSCC (2021)
Keyphrases
- neural network
- auto associative
- associative memory
- digital signal processors
- compute intensive
- artificial neural networks
- low cost
- limited memory
- back propagation
- feed forward
- bp neural network
- low memory
- parallel implementation
- bayesian networks
- belief networks
- self organizing maps
- genetic algorithm
- neural network is trained
- probabilistic inference
- highly scalable
- memory size
- memory efficient
- general purpose
- graphical models
- loopy belief propagation
- field programmable gate array
- feed forward neural networks
- memory requirements
- multilayer perceptron
- recurrent neural networks