All-You-Can-Fit 8-Bit Flexible Floating-Point Format for Accurate and Memory-Efficient Inference of Deep Neural Networks.
Cheng-Wei HuangTim-Wei ChenJuinn-Dar HuangPublished in: CoRR (2021)
Keyphrases
- floating point
- efficient inference
- neural network
- probabilistic inference
- floating point arithmetic
- instruction set
- fixed point
- conditional random fields
- markov random field
- fully connected
- human pose estimation
- exact inference
- artificial neural networks
- high quality
- graphical models
- image processing
- structured prediction
- computer vision
- high resolution
- scale free
- hidden variables
- influence diagrams
- graph structure
- multilayer perceptron
- back propagation