Receptive field
This is student work which has not yet been approved as correct by the instructor
Introduction[edit]
Receptive fields used in an artificial neural networks are similar to the biological explanation. They are a sensory neuron where a prompt, or stimulation, changes the action of the neuron.[2] In other words, the fields gather information to be used by the neurons. They are used in relation to convolutional neural networks (CNNs). The neurons are arranged in a 3-D array. [3]
How does it work or a deeper look[edit]
The receptive field gathers information and sends the information to the neurons in the CNNs. Rather than a more traditional neural network where every input neuron is linked to every other neuron at the first level, Receptive fields allow for "pre-processing" of just a certain grid (or part) of the input image.
Examples[edit]
The Gaussian receptive fields are being used in test on cars to allow them to be detected. [4]
Pictures, diagrams[edit]
- An image showing the arrangement of neurons in the CNN.
- An image showing the receptive field and the CNN.
External links[edit]
- http://www.inase.org/library/2015/books/bypaper/MCSI/MCSI-42.pdf
- https://en.wikipedia.org/wiki/Receptive_field
- https://en.wikipedia.org/wiki/Receptive_field#In_the_context_of_neural_networks
- https://en.wikipedia.org/wiki/Convolutional_neural_network