A 28nm 128TFLOPS/W Computing-In-Memory Engine Supporting One-Shot Floating-Point NN Inference and On-Device Fine-Tuning for Edge AI.
Haikang DiaoHaoyang LuoJiahao SongBocheng XuRunsheng WangYuan WangXiyuan TangPublished in: CICC (2024)
Keyphrases
- floating point
- fine tuning
- neural network
- fixed point
- artificial intelligence
- nearest neighbor
- square root
- viable alternative
- inference engine
- memory bandwidth
- fine tune
- expert systems
- fine tuned
- interval arithmetic
- memory requirements
- sparse matrices
- edge detection
- fast fourier transform
- instruction set
- floating point arithmetic
- knn
- image segmentation
- bayesian networks
- pairwise