Information Technology Reference
In-Depth Information
Fig. 3.
Trackers matching trends T1 to T7 with no memory feedback
It should be noted however that due to apoptosis during run 4 of the experi-
ment a number of important trackers were eliminated before they had a chance
to bind. This resulted in the TEA failing to detect 6 of the 7 available trends in
this run. Omitting this unusual occurrence from our analysis would have boosted
the current 80% detection rate to 87%.
Fig. 4.
Total Tracker and total matching tracker populations with memory feedback
Figures 4 and 5 show the total tracker population levels and tracker popula-
tions that match the specific trends T1 to T7. Figure 4 shows a more pronounced
secondary response to A2 compared to that in Figure 2, with the maximum
population rising to 191 trackers compared to that of 44 in experiment 1. Look-
ing at the population of trackers that map to specific trends (Figure 5) we see
evidence of stronger responses to the trends in A2, as seen in Figure 3. Thus
knowledge of the trends seen from A1's presentation have improved the TEA's
recognition of new, novel trends that have some association with those previously
seen. This leads to the 43.6% improvement in trend detection.
