Manifold-tiling Localized Receptive Fields are Optimal in Similarity-preserving Neural Networks.
Anirvan M. SenguptaCengiz PehlevanMariano TepperAlexander GenkinDmitri B. ChklovskiiPublished in: NeurIPS (2018)
Keyphrases
- receptive fields
- neural network
- hidden layer
- biologically inspired
- natural images
- visual cortex
- fuzzy logic
- pattern recognition
- similarity measure
- back propagation
- neural nets
- space time
- feed forward
- neural network model
- multilayer perceptron
- saliency map
- data sets
- higher order
- training set
- image segmentation
- primary visual cortex