Efficient Fixed/Floating-Point Merged Mixed-Precision Multiply-Accumulate Unit for Deep Learning Processors.

Hao Zhang Hyuk Jae Lee Seok-Bum Ko

Published in: ISCAS (2018)

Keyphrases

floating point
deep learning
instruction set
sparse matrices
fixed point
unsupervised learning
probabilistic model
parallel algorithm
floating point arithmetic
object detection
weakly supervised
processing units
unsupervised feature learning