Login / Signup
SP-PIM: A 22.41TFLOPS/W, 8.81Epochs/Sec Super-Pipelined Processing-In-Memory Accelerator with Local Error Prediction for On-Device Learning.
Jung-Hoon Kim
Jaehoon Heo
Wontak Han
Jaeuk Kim
Joo-Young Kim
Published in:
VLSI Technology and Circuits (2023)
Keyphrases
</>
active learning
supervised learning
knowledge acquisition
learning systems
real time
learning algorithm
learning tasks
computational power
cognitive load
human memory
machine learning
data streams
unsupervised learning
main memory
prediction error
compute intensive