Information Technology Reference
In-Depth Information
Tikk, D., Kóczy, L. T., & Gedeon, T. D. (2001).
Universal approximation and its limits in soft comput-
ing techniques. An overview
(Research Working Paper No. RWP-IT-02-2001). Perth (W.A.): Murdoch
University.
Vester, F. (1988). The biocybernetic approach as a basis for planning our environment.
Systems Practice
,
4
, 399-413.
Weber, J., & Schäffer, U. (2000).
Balanced Scorecard und Controlling. Implementierung - Nutzen
für Manager und Controller - Erfahrungen in deutschen Unternehmen
(3
rd
ed.). Wiesbaden: Dr. Th.
Gabler Verlag. (in German)
endnotes
1
If the two Humean properties of causality would be sufficient to explain causal structures, a bad
weather forecast would be the regarded as the cause for bad weather.
2
An α -error or type-1-error denotes a cause-and-effect relation between two variables in a causal
strategy model although the corresponding time-series are not causal. On the contrary the term
β- or type-2-error is used if there exists a causal relation in reality (i.e. between the time-series of
two variables but it is not represented in the causal strategy model.
3
The computation of the autocorrelation of a variable
X
and a given order
n
corresponds to the one
of the cross correlation
X X
n
. Hence it will be abbreviated by the symbol
( )
( )
X
n
for the rest of
this chapter.
4
IC = Inductive Causation
5
The transitive closure is the set of all direct and indirect successors of a node within a directed
graph (e.g. a causal strategy model).
6
A validation data set is a subset of the empirical data which is not used to train the ANN
11
∑
7
e.g.:
net
=
in
+
w
⋅
out
e
e
h e
,
h
t
−
2
t
−
2
i t
−
2
i
i
=
9
8
The statistical significance of this statement has been validated by testing the group average of
the squared errors for both techniques using variance analysis.
Search WWH ::
Custom Search