Feed-Forward On-Edge Fine-tuning Using Static Synthetic Gradient Modules.
Robby NevenMarian VerhelstTinne TuytelaarsToon GoedeméPublished in: CoRR (2020)
Keyphrases
- feed forward
- fine tuning
- edge detection
- back propagation
- viable alternative
- image gradient
- neural nets
- recurrent neural networks
- artificial neural networks
- gradient direction
- fine tune
- gradient magnitude
- edge orientation
- step edges
- biologically plausible
- gradient information
- orientation histogram
- neural network
- hidden layer
- edge detector
- gradient field
- zero crossing
- visual cortex
- feed forward neural networks
- activation function
- edge information
- neural architecture
- artificial neural
- neuron model
- error back propagation
- primate visual cortex
- fine tuned
- recurrent networks
- edge map
- image derivatives
- mumford shah model
- scale space
- knowledge base
- edge direction
- weighted graph
- bp neural network
- data mining