Approximate computation based on NAND-SPIN MRAM for CNN on-chip training.
Zhengyi HouLuyao ShiBi WangZhaohao WangPublished in: NANOARCH (2022)
Keyphrases
- random access memory
- efficient computation
- design considerations
- low power consumption
- low cost
- cellular neural networks
- training set
- flash memory
- high speed
- training process
- training algorithm
- circuit design
- supervised learning
- exact computation
- database
- power consumption
- test set
- training examples
- training samples
- pre computed
- numerical integration
- analog vlsi