Questo sito utilizza cookie di terze parti per raccogliere informazioni statistiche.
scorrendo questa pagina, cliccando su un link o proseguendo la navigazione in altra maniera, acconsenti all' uso dei cookie.

Artificial Sciences




With Artificial Science we mean those sciences for which the comprehension of the natural and /or cultural processes is obtained through the re-creation of those same processes through automatic models.
In the A.S. the computer is what the handwriting is for the natural language: the A.S are made of formal algebras for the generation of artificial models (structures and processes), in the same way in which the natural languages are constituted by a semantics, a syntax and a pragmatics for the generation of the texts.

In natural languages the writing is the achievement of the independence from the time of the word, through the space; in the A.S the computer is the achievement of the model of the independence from the subject, through the automation and/or the action.
Just as through the writing a natural language can generate cultural objects that were unthinkable before the writing (novels, texts of laws, manuals) in the same way the A.S. through the computer can generate automatic models particularly complex.
Thus, natural languages and Artificial Science, without the writing and the computer, remain limited. But a writing that is not based upon a natural language or an automatic model that is not generated by a formal algebra is a set of scribbles.



In the A.S. the comprehension of any natural and /or cultural process occurs in a way proportional to the capacity to recreate that process of the automatic artificial model. The more the comparison between original process and generated model gives a positive outcome, the more it is probable that the artificial model has made clear the working rules of the original process.
However, this comparison cannot be made in a naïve way. Sophisticated analysis instruments are necessary to carry out a reliable comparison between original process and artificial model.
The majority of instruments of analysis useful for this comparison consists in comparing the dynamics of the original process with those of the artificial model when the respective boundary conditions change.
In summary it might be said:
- the more with the change of the boundary conditions we obtain variety of dynamics of responses both in the original process and in the artificial model;
- the more this dynamics between original process and artificial model are homologous, then
- the more it is probable that the artificial model is a good explanation of the original process.

We propose a taxonomic tree for the characterization of disciplines that, through the Natural Computation and the Classic Computer Science, compose the system of the Artificial Science.





Natural computation
With Natural Computation is meant that part of the Artificial Sciences (A.S.) that tries to built automatic model of Natural and /or cultural processes through the local interaction of micro processes which are not isomorphous to the original process.
Thus, in the N.C. it is assumed that any process is the result, more or less contingent, of more elementary processes that tend to self-organise in time and in space and that none of these micro processes is for itself informative about the function that it will take with respect to others, neither of the global process it will be part of.
This computational Philosophy, not very economical for the creation of simple models, can be used effectively to create any type of process or model that draws its inspiration from complex processes, namely to processes for which the classical philosophies have found remarkable disadvantages. This is the reason for N.C. the analysis and the generation of highly not linear artificial model.

The N.C. deals, actually, with the construction of artificial models that simulate the complexity of the natural and /or cultural processes not through rules, but through bonds that according to the space and the time through which the process takes shape, creates autonomously a set of contingent and approximated rules.
The NC does not try to recreate natural and /or cultural processes analysing the rules through which they should work and formalising these rules in an artificial model. On the contrary, N.C. tries to recreate natural and /or cultural processes constructing artificial models able to recreate dynamically local rules susceptible of change in accordance with the same process.
The bonds that allow the N.C. models to generate rules dynamically look like Kantian transcendental rules: it deals with rules that establish conditions of possibility of other rules. In the N.C. the dynamics such as the learning to learn is implicit in the same artificial models, while in the Classical Computation it needs additional rules.
The graph shows in a more detailed way the formalization, the automation and the comparison between natural and /or cultural processes and Automatic Artificial Models from two different points of view (Classical Computation and Natural Computation). Each point of view can be seen as a cycle that can repeat more times. From here it can be assumed that the human scientific process that characterizes both cycles look more like the one on the Natural Computation rather than like the one of the Classical Computation.




Descriptive Systems
With Descriptive Systems (D.S.) are meant those disciplines that have developed, more or less intentionally, formal algebras that turned particularly effective in elaborating proper bonds of functioning of artificial models generated inside N.C.


Generative Systems
With Generative Systems (G.S.) are meant those theories of N.C. that explicitly deal with formal algebras oriented to generate artificial models of natural and /or cultural processes through bonds that create dynamic rules in time and space.


Phisycal Systems
With Physical Systems are obtained by gathering those theories of Natural Computation whose generative algebras create artificial model comparable to material and /or cultural processes, only when the artificial model reaches certain evolutive phases (like limit cycle). While not necessarily the route through which the bonds generate the model is itself a model of the process of the origin process. In shorty, in these systems the time of generation of the model is not necessarily an artificial model of evolution of the time of the process.


Sistemi adattivi
With Adaptive Systems (S.A.) are meant those theories of N.C. whose generative algebras create artificial models of natural and /or cultural processes, whose process of the birth of the process is itself an artificial model comparable with the birth of the process of origin. Thus, it deals with theories that assume the time of emergence of the model as a formal model of the time of the process itself.
In short: for these theories each phase of artificial generation is a model comparable to a natural and /or cultural process.
The Adaptive Artificial Systems include:

  • Learning Systems (Artificial Neural Networks - ANNs): are meant those algorithms for the information processing that allow to rebuilt, in a particularly effective way, the approximate rules that establish a connection a certain set of “explicative” data for the considered problem (input), with a set of data (Output) for which a correct prediction or reproduction in conditions of informative incompleteness is required.
  • Evolutivi (Evolutive Systems): that change with time their architecture and their functions in order to adapt to the environment in which they are inserted. The evolution of a genotype from a time ti to a time t(i+n) is a good example of the time evolution of the architecture and of the functions of an adaptive systems. In the evolutive systems different levels of change of the architecture and of the system functions can be considered:

    a) time evolution of the type of states that configure the possibilities of the system. The families of the Genetic Algorithms belong to this sector;

    b) time Evolution of the type of states and transformation among states that configure the evolutive metric of the System. In this case, for each system there is also the evolution of the functions with which it generates its own states. In this sector the Genetic Programming developed.

    c) time Evolution also of the optimal conditions of the relationship between system and environment. It deals with generating systems that live in a variable environment. Consequently, their function of adaptation to the environment change itself with time (fitness), evolving, forcing all the systems to an unceasing change of evolutive strategy. The Evolutive Strategy deals with this field.

    d) time Evolution of Complex Organisms each of which not only evolves its state, changes its operators between states and changes its strategies as a function of the environmental variations, but it designs strategies, it operates predictions about itself and about the others, it tries to establish evolutive rules different from the existing ones. This sector still exists only partly as research field. It concerns the Evolution of Agents and the labels under which, perhaps, it should be studied in depth the Organism programming.


Semeion Centro Ricerche di Scienza della Comunicazione
Via Sersale 117, 00128 Roma - Tel. 0650652350, Fax 065060064 - C.F. 06911530589 P.IVA 01644611004

Joomla templates by a4joomla