Accuracy to Throughput Trade-Offs for Reduced Precision Neural Networks on Reconfigurable Logic.
Jiang SuNicholas J. FraserGiulio GambardellaMichaela BlottGianluca DurelliDavid B. ThomasPhilip Heng Wai LeongPeter Y. K. CheungPublished in: ARC (2018)
Keyphrases
- neural network
- trade off
- precision and recall
- high accuracy
- high precision
- pattern recognition
- precision recall
- artificial neural networks
- computational cost
- accuracy rate
- general purpose
- genetic algorithm
- classification accuracy
- error rate
- fault diagnosis
- low cost
- response time
- training process
- classical logic
- learning algorithm