Training and Inference using Approximate Floating-Point Arithmetic for Energy Efficient Spiking Neural Network Processors.
Myeongjin KwakJungwon LeeHyoju SeoMingyu SungYongtae KimPublished in: ICEIC (2021)
Keyphrases
- energy efficient
- wireless sensor networks
- energy consumption
- spiking neural networks
- sensor networks
- biologically inspired
- floating point
- floating point arithmetic
- instruction set
- routing protocol
- parallel algorithm
- base station
- training algorithm
- energy efficiency
- parallel processing
- training process
- parallel computing
- artificial neural networks
- neural network
- real time
- sensor nodes
- sensor data
- learning rules
- biologically plausible
- artificial intelligence