Author:
Andrew Schumann
Affiliation:
University of Information Technology and Management in Rzeszow, Poland
Abstract:
Artificial actin filament networks (AAFN) we are going to design employ
transmitting information by means of building and rebuilding actin filaments
in responses to dynamics of intra-cellular and extra-cellular stimuli. Actin is one
of the most important proteins responsible for a reaction of cells to different stresses
including internal and external. In responses to stimuli, filaments can be organized
as complex networks of different forms: (i) unstable bunches forming a
filament wave, (ii) trees supporting the membrane or changing it to pseudopodia,
(iii) stable bunches for transmitting mechanical forces. For designing protein
robots, we are interested in managing reactions of AAFN by modeling external
stimuli. There are the following two basic types of all the external stimuli: (i)
attractants (which stimulate the directed movement towards signals by means of
building and rebuilding actin filament waves towards the signals) and (ii) repellents
(which stimulate the directed moveme
nt away from signals by means of
building and rebuilding actin filament waves in the opposite direction). In this
model, we have inputs as different stresses and outputs as assemblies and disassemblies
of actin filament waves. Thus, under different external conditions we
observe different dynamics in changing the outlook of AAFN. In this way AAFN
can be represented as a basic mechanism of the primitive protein robots, in which
actin filament waves are regarded as processors. These processors can appear in
one setting and they can disappear in other settings. The system of AAFN is
much more complex and effective, than artificial neural networks studied well.
The point is that in artificial neural networks, there is a fixed number of processors
(called neurons) and there are many connections among neurons which
are changeable. In the AAFN, both processors (called actin filament waves) and
all the connections among them are changeable simultaneously. As a result, the
AAFN can solve much more tasks, than artificial neural networks. Now there is
no mathematical theory of AAFN. Nevertheless, if it is possible to create an artificial
protein broth which will be a robot solving the complex of various tasks
(learning, orientation in space, moving, decision making about own transitions,
etc.), then this broth will consist of actin filaments controlled by us. In the project
for the first time we are going to propose a mathematical theory of AAFN
which can be used for designing communication networks organized among mobile
agents, communicated in large groups which can joint higher-order groups,
as well.
(More)