CAxCNN: Towards the Use of Canonic Sign Digit Based Approximation for Hardware-Friendly Convolutional Neural Networks.
Mohsin RiazRehan HafizSalman Abdul KhaliqMuhammad FaisalHafiz Talha IqbalMohsen AliMuhammad ShafiquePublished in: IEEE Access (2020)
Keyphrases
- convolutional neural networks
- hardware and software
- low cost
- convolutional network
- hardware implementation
- approximation algorithms
- computer systems
- closed form
- real time
- error bounds
- approximation error
- massively parallel
- approximation methods
- multiresolution
- recognition rate
- embedded systems
- lower bound
- image processing
- continuous functions
- vlsi implementation
- information systems
- neural network