Introducing time dependence requires the activation to be defined by its rate of change dtda, the derivative. The input weighted sum is then
uj=j∑wij⋅xi
and the activation is
y=dtda=−η a+β u
where η and β are positive constants. −η a gives rise to activation decay. βu represents the input from other neurons.
Decay effect is summarised as follows:
- a>0⟶dtda<0: a decreases
- a<0⟶dtda>0: a increases
The neuron will reach equilibrium when dtda=0. That is, when u=ηβu.