Occasionally the word gate is used here and in biological descriptions of synapse
processes to describe the process of modulating, or gating, an input value on a synapse by a weight
or connection strength value. The resulting value is a weight-determined portion of the input-stimulus value. This is sometimes referred to as gating because conceptually, the weight value determines the magnitude of the signal value that passes, like adjusting the opening and closing of a gate would determine the amount of substance that passes through it.
The biological processes responsible for modulating the connection strength to a neuron
are often referred to as gate, or gating, processes. In biological synapses, such gating modulates the strengths of connected signals by permitting varying amounts of ion
s to pass in and out of the neuron. It is conceptually based on how far (or more accurately for biological cells, how often) the gate is figuratively opened or closed (see, Irwin Levitan
, Leonard Kaczmarek
, "The Neuron: Cell And Molecular Biology
", 2002, Oxford University Press, ISBN: 0-19-514523-2).
In the case of the biological synapse
, the type of the synapse determines whether the amount of signal gated will have an inhibitory
, or excitatory
effect on the neuron.
Purely mathematical simulations usually refer to this as a multiplication because it has become customary to talk about such neural network
models in terms of functions performed using floating point arithmetic.
In these conventional neural networks
the sign of the weight usually serves as a type-signifier, which determines whether the connection is inhibitory or excitatory. A negative weight value specifies an inhibitory synapse, and a positive weight value specifies an excitatory synapse. In both cases, the absolute value represents the connection strength
between the pre- and post-synaptic