Analysis of a Growth Model for Idiotypic Networks - Artificial Immune Systems - page 72

Fig. 2. Cells in the network define tolerated and non-tolerated zones in the space.
Blank areas are tolerant to any added cells; the darker the shading, the more reactive
the region to added cells.

5.1 Overview of Emergent Network Structure

Figure 1 shows the evolution of two networks using a homogeneous and het-
erogeneous growth model. In both cases following a rapid growth period, the
networks rapidly stabilize to a relatively constant size. For the three values of
upper limit U investigated, the heterogeneous growth model tends to produce
larger networks. Previous work with homogeneous networks in [10,9] showed
that the resultant emergent networks segregate the 2D space into a number of
regions which define tolerant and non-tolerant regions of the space (without any
need to pre-label cells as a particular type). The results obtained here with a
heterogeneous growth model concur exactly with the homogeneous case. In both
models, visualisation of the networks in the 2D space shows that cells that are