For the ART2 learning function ` ART2` there are various parameters
to specify. Here is a list of all parameters known from the theory:

Vigilance parameter. (first parameter of the learning and update function). is defined on the interval For some reason, described in [Her92] only the following interval makes sense:

aStrength of the influence of the lower level in F by the middle level. (second parameter of the learning and update function). Parameteradefines the importance of the expection of F , propagated to F : Normally a value of is chosen to assure quick stabilization in F .

bStrength of the influence of the middle level in F by the upper level. (third parameter of the learning and update function). For parameterbthings are similar to parametera. A high value forbis even more important, because otherwise the network could become instable ([CG87b]).

cPart of the length of vectorp(unitsp...p) used to compute the error. (fourth parameter of the learning and update function). Choosecwithin0 < c < 1.

dOutput value of the F winner unit. You won't have to passdtoART2, because this parameter is already needed for initialization. So you have to enter the value, when initializing the network (see subsection on the initialization function). Choosedwithin0 < d < 1. The parameterscanddare dependent on each other. For reasons of quick stabilizationcshould be chosen as follows: . On the other handcanddhave to fit the following condition:

ePrevents from division by zero. Since this parameter does not help to solve essential problems, it is implemented as a fix value within the SNNS source code.Kind of threshold. For the activation values of the units

xandqonly have small influence (if any) on the middle level of F . The output functionfof the unitsxandqtakes as its parameter. Since this noise function is continuously differentiable, it is calledOut_ART2_Noise_ ContDiffin SNNS. Alternatively a piecewise linear output function may be used. In SNNS the name of this function isOut_ART2_Noise_PLin. Choose within

To train an ART2 network, make sure, you have chosen the learning
function ` ART2`. As a first step initialize the network with the
initialization function ` ART2_Weights` described above. Then set
the five parameters , **a**, **b**, **c** and , in the
parameter windows 1 to 5 in both the ` LEARN` and ` UPDATE`
lines of the control panel. Example values are 0.9, 10.0, 10.0, 0.1, and
0.0. Then select the number of learning cycles, and finally use the buttons
and to train a single pattern or all patterns
at a time, respectively.

Niels.Mache@informatik.uni-stuttgart.de

Tue Nov 28 10:30:44 MET 1995