Compute-In-Memory SRAM Macro with Analog-Digital Hybrid Computing for Deep Neural Network Inference.
Sangsu JeongJuyoung OhDongsuk JeonPublished in: CICC (2024)
Keyphrases
- neural network
- auto associative
- power consumption
- hybrid intelligent
- associative memory
- random access memory
- artificial neural networks
- mixed signal
- low power
- efficiently computing
- analog circuits
- signal processing
- memory requirements
- data conversion
- analog to digital converter
- delta sigma
- probabilistic inference
- low cost
- back propagation
- bayesian inference
- bayesian networks
- belief nets
- self organizing maps
- memory space
- memory usage
- feed forward
- main memory
- radial basis function
- fault diagnosis
- pattern recognition
- neural network is trained
- efficient computation
- sigma delta
- dynamic random access memory
- genetic algorithm