Thus in practice only the winning unit or best matching unit
dcis updated. In (lb), the parameter CY,with 0 5 a, 5 1
is the learning rate which is usually a small real number, or
it starts from a reasonable initial value and then reduces to
zero in some way [14], [15], e.g., in the way used in the
Robbins-Monro stochastic approximation procedure [21]. The
explicit dependence of CY, on time is not shown above.
Note: Equations (la) and (lb) are often called the WinnerTake-All rule. In addition, there are also some other versions
of the classical CL [26]. However, the basic idea is the same;
the weight vector of the neural unit which tunes to an input
most strongly is adjusted most strongly to tune to the input
even stronger
Thus in practice only the winning unit or best matching unitdcis updated. In (lb), the parameter CY,with 0 5 a, 5 1is the learning rate which is usually a small real number, orit starts from a reasonable initial value and then reduces tozero in some way [14], [15], e.g., in the way used in theRobbins-Monro stochastic approximation procedure [21]. Theexplicit dependence of CY, on time is not shown above.Note: Equations (la) and (lb) are often called the WinnerTake-All rule. In addition, there are also some other versionsof the classical CL [26]. However, the basic idea is the same;the weight vector of the neural unit which tunes to an inputmost strongly is adjusted most strongly to tune to the inputeven stronger
การแปล กรุณารอสักครู่..
