Image-Classifier Deep Convolutional Neural Network Training by 9-bit Dedicated Hardware to Realize Validation Accuracy and Energy Efficiency Superior to the Half Precision Floating Point Format.
Shin-ichi O'UchiHiroshi FuketaTsutomu IkegamiWakana NogamiTakashi MatsukawaTomohiro KudohRyousei TakanoPublished in: ISCAS (2018)
Keyphrases
- floating point
- energy efficiency
- convolutional neural network
- dedicated hardware
- image segmentation
- instruction set
- fixed point
- energy consumption
- power consumption
- sensor networks
- training set
- wireless sensor networks
- routing protocol
- floating point arithmetic
- machine learning
- data center
- ambient intelligence
- low cost
- data structure
- image processing