inhibition

Terms from Artificial Intelligence: humans at the heart of algorithms

Page numbers are for draft copy at present; they will be replaced with correct numbers when final book is formatted. Chapter numbers are correct and will not change now.

In a neural network and indeed animal nervous ssytems, one neoron inhibits another, if the activation of the first reduces the activation of the second. In a simple neural network this is represented by a negative weight between the neurons. In a feedforward network this affect only operate forward from one layer to the next, but in other forms of neural network there may be mutual inhibition between neurons in the same layer (also known as lateral inhibition) or even back-and-forth between layers (if there are feeedback connections between the layers). Inhibition is critical to be able to represent negatives (such as NOT having legs) and mutual inhibition can help disambiguation and classification.

Defined on page 123

Used on Chap. 6: page 123

Also known as inhibit