Natural Computing Paradigms and Emergent Computation - Systematic Design for Emergence in Cellular Nonlinear Networks - page 4

the nonlinear ordinary differential equation describing the cell. Subdomains of the
cell parameters were found (called “edge of chaos” domains) where emergent
behavior and complexity is likely to occur. The principle of local activity is a first
step towards the establishment of a new science of designing for emergence.
1.2 Open Problems and Book Description
Looking around in nature it is clear that above principles apply for various systems
ranging from living beings, economic and social networks, international relation-
ships, computer networks, etc. Moreover, at a critical observation all functional
architectures designed so far by information engineers are natural computing
architectures. Indeed, any computer is built upon an hierarchy of functional mod-
ules where at the basic level one finds the basic cell to be the locally active and
nonlinear transistor (or any other device exhibiting local activity). Various func-
tions were discovered through a process of mental evolution, which gives the
proper interconnection between such basic cells. A further level of hierarchy was
the creation of logic gates, then the creation of memory cells, arithmetic units etc,
up to the level where some flexibility was added to this construction through what
is today called software , i.e. a flexible description of how different cells should be
connected to perform a certain task. Algorithms then are also designed as inter-
connections of basic “software” cells called instructions in the computer science
parlance.
However, designing such systems is based on description rather than emer-
gence. Description can be regarded as a process of transferring human knowledge
into a model of existing cells. In other words we have the cells (i.e. the transistors
or the logic gates or more sophisticated modules, etc.) and we describe how to
interconnect them such that the resulting system will manifest a desired function-
ality. The typical design process starts with a specification of the end function and
proceeds with a given set of rules or procedures (e.g. methods for drawing sche-
matics or programming in software) to interconnect the basic cells available in
a given technology to produce the expected result. Although description is itself
an emergent process whenever novel functionality has to be described, it turns out
that several challenging problems in information technology are not enough
satisfactory solved using the “classic design” method.
A list of problems includes but is not limited to such functions as:
x Self-repair and self-reproduction of complex functional architectures
x Voice recognition (speaker identification and message identification)
x Gesture recognition
x Visual field abstraction (i.e. generating a brief description from a visual scene)
x NP-complete problems optimization and decision making
x Orientation in a complex environment
x Creative thinking and problem solving.