Reconstructed spatial receptive field structures by reverse correlation technique explains the visual feature selectivity of units in deep convolutional neural networks.
Yoshiyuki R. ShiraishiHiromichi SatoTakahisa M. SanadaTomoyuki NaitoPublished in: CoRR (2021)
Keyphrases
- visual features
- receptive fields
- convolutional neural networks
- visual information
- space time
- biologically inspired
- saliency map
- image classification
- natural images
- image retrieval
- image representation
- visual cortex
- low level features
- spatio temporal
- primary visual cortex
- keywords
- hidden layer
- low level
- spatial information
- input space
- learning algorithm
- image processing
- key frames
- high dimensional
- machine learning