Information Technology Reference
In-Depth Information
Fig. 12.5
The results of running the model with all the information in a given series DL1 and 2 as
shown in a or limited to a window WL1 and 2 as shown in b Fig.
12.2
The running probability (i.e. y-axis) is the weighted average of
four
examples of
a test sequence using a window size of
seven
. The selection of window size seven
was made after running the model for various window sizes. We found that for our
examples, seven was the optimal number for any significant differences to be detected
in the values of the running probability. However, we must stress that seven is not an
ideal number for all sizes of window. Different samples may need to be smaller or
bigger in order for a change in the values of running probabilities to be significant.
Note here that Eysenck only used an average of four numbers in his sequences for
IQ tests.
The behavior of the system is shown by the graphs. This graphs indicate a sim-
ilar trend for both experiments. The generating hypothesis shows an increase in its
probabilistic values over time while others tend to collapse or be unstable. There
are instances where two hypotheses are competing against each other where both
their conditional probability values are on the increase. This phenomenon suggests
that both hypotheses are actually two different manifestation of the same thing. For
example, some polynomial functions can be represented as a periodic functions and
vice-versa.
The results showed that the running conditional probability has the potential for
limiting the exploration for a hypothesis. As we continue to assess a sequence, the
probability value of some concepts increases while others decreases. A consistent
decrease in probability value indicate that a particular concept should be abandoned
from further consideration.
