Vitorino Ramos: on the implicit and on the artificial

VITORINO RAMOSΠ
aLife Art Architecture Lab
+ CVRM / IST Technical Univ. of Lisbon

On the Implicit and on the Artificial⊕
Morphogenesis and Emergent Aesthetics in
Autonomous Collective Systems
“[…]QUESTION_HUMAN > If Control’s control is absolute, why does Control need
to control?
ANSWER_CONTROL > Control…, needs time.
QUESTION_HUMAN > Is Control controlled by his need to control ?
ANSWER_CONTROL > Yes.
QUESTION_HUMAN > Why is Control need Humans, has you call them ?
ANSWER_CONTROL > Wait ! Wait…! Time are lending me…;
Death needs time like a Junkie… needs Junk.
QUESTION_HUMAN > And what does Death need time for ?
ANSWER_CONTROL > The answer is so simple ! Death needs time for
what it kills to grow in ! […]”,
in Dead City Radio, William S. Burroughs / John Cale , 1990.

Imagine a “machine” where there is no pre-commitment to any particular
representational scheme: the desired behaviour is distributed and roughly specified
simultaneously among many parts, but there is minimal specification of the mechanism
required to generate that behaviour, i.e. the global behaviour evolves from the many
relations of multiple simple behaviours . A machine that lives to and from/with Synergy.
An artificial super-organism that avoids specific constraints and emerges within
multiple low-level implicit bio -inspired mechanisms.
The emergence of complex behaviour in any system consisting of interacting simple
elements is among the most fascinating phenomena of our world. Examples can be found in
almost every field of today’s scientific interest, ranging from coherent pattern formation in
physical and chemical systems, to the motion of swarms of animals in biology, and the
behaviour of social groups. In the life and social sciences, one is usually convinced that the
evolution of social systems is determined by numerous factors, difficult to grasp, such as
cultural, sociological, economic, political, ecological, etc. However, in recent years, the
development of the interdisciplinary fields “science of complexity”, along with “artif icial
life” (aLife), has lead to the insight, that complex dynamic processes may also result from
simple interactions. Moreover, at a certain level of abstraction, one can also find many
common features between complex structures in very different fields.
Francis Heylighen, mentor of the Principia Cybernetic Project, an international
organization (PCP, http://pespmc1.vub.ac.be/) points precisely to this paradigm shift, with a
remarkable historical perspective, namely in what concerns the view within the social
sciences, using biology as a metaphor, and more recently those from complexity science. In
“The Global Superorganism: an Evolutionary-Cybernetic Model of the Emerging Network
Society” (Journal of Social and Evolutionary Systems, 2001) he writes:

Π

http://alfa.ist.utl.pt/~cvrm/staff/vramos
in ARCHITOPIA Book / Catalogue, Art, Architecture and Science, J.L. Maubant and L.
Moura (Eds.), pp. 25-57, Ministério da Ciência e Tecnologia, Feb. 2002.
⊕

RAMOS, V., On the Implicit and on the Artificial

2

On the right, a synthetic test image composed by small squares of different sizes, but
perceived by us (humans) as a big cross, probably due to our inner perceptual grouping
Gestalt laws (Wertheimer, 1910). On the left, the perceived cross by an artificial ant colony as
a function of the spatial distribution of pheromone (Swarm Cognitive Map), at T=1000
(Ramos, 1998-2000).

[…] It is an old idea that society is in a number of respects similar to an organism, a
living system with its cells, metabolic circuits and systems. As an example, the army
functions like an immune system, protecting the organism from invaders, while the
government functions like the brain, steering the whole and making decisions. In this
metaphor, different organizations or institutions play the role of organs, each fulfilling its
particular function in keeping the system alive, an idea that can be traced back at least as far
as Aristotle, being a major inspiration for the founding fathers of sociology, such as Comte,
Durkheim and especially Spencer […]
Then at this point, Heylighen stresses the importance of recognizing the underlying
component of complexity in nature, a bottom-up view common to the field of Artificial Life:
[…] The organismic view of society has much less appeal to contemporary theorists.
Their models of society are much more interactive, open-ended, and non-deterministic than
those of earlier sociologists, and they have learned to recognize the intrinsic complexity and
unpredictability of society. The static, centralized, hierarchical structure with its rigid division
of labor that seems to underlie the older organismic models appears poorly suited for
understanding the intricacies of our fast-evolving society. Moreover, a vision of society where
individuals are merely little cells subordinated to a collective system has unpleasant
connotations to the totalitarian states created in the last century. As a result, the organismic
model is at present generally discredited in sociology […]
Similarly, biology has traditionally started at the top, viewing a living organism as a
complex biochemical machine, and has worked analytically down from there through the
hierarchy of biological organization – decomposing a living organism into organs, tissues,
cells, organelles, and finally molecules – in its pursuit of the mechanisms of life. Analysis
means ‘the separation of an intellectual or substantial whole into constituents for individual
study’ (that is, by top-down reductionist approaches). By composing our individual
understandings of the dissected component parts of living organisms, traditional biology has
provided us with a broad picture of the mechanics of life on Earth.
In the meantime, however, new scientific developments have done away with rigid,
mechanistic views of organisms (Heylighen). As pointed by Langton, there is more to life
than mechanics – there is also dynamics. Life depends critically on principles of dynamical
self-organization that have remained largely untouched by traditional analytic methods. There


3

is a simple explanation for this – these self-organized dynamics are fundamentally non-linear
phenomena, and non-linear phenomena in general depend critically on the interactions
between parts: they necessarily disappear when parts are treated in isolation from one another,
which is the basis for any analytic method. Rather, non-linear phenomena are most
appropriately treated by a synthetic approach, where synthesis means “the combining of
separate elements or substances to form a coherent whole’. In non-linear systems, the parts
must be treated in each other’s presence, rather than independently from one another, because
they behave very differently in each other’s presence than we would expect from a study of
the parts in isolation. As suggested by Langton, the key concept in aLife is emergent
behaviour. Natural life emerges out of the organised interactions of a great number of
nonliving molecules, with no global controller responsible for the behaviour of every part.
Rather, every part is a behaviour itself, and life is the behaviour that emerges from out of all
of the local interactions among individual behaviours. It is this bottom-up, distributed local
determination that aLife employs in its primary methodological approach to the generation of
lifelike behaviours. Of course, there is no universally agreed definition of life. The concept
covers a cluster of properties, most of which are themselves philosophically problematic: selforganization, emergence, autonomy, growth, development, reproduction, evolution,
adaptation, responsiveness, and metabolism. Scientists differ about the relative importance of
these properties, although it is generally agreed that the possession of most (not necessarily
all) of them suffices for something to be regarded as alive. One common concept however is
shared by all: complex behaviour can emerge from a system consisting of interacting simple
elements. Once more, Heylighen stresses biology as an example :
[…] when studying living systems, biologists no longer focus nowadays on the static
structures of their anatomy, but on the multitude of interacting processes that allow the
organism to adapt to an ever changing environment. Recently, the variety of ideas and
methods that is commonly grouped under the head of the sciences of complexity , has led to
understanding that artificial organisms can be self-organizing, adaptive systems. Most
processes in such systems are decentralized, non-deterministic and in constant flux. They
thrive on noise, chaos and creativity. Their collective intelligence emerges out of the free
interactions between individually autonomous components […]
In fact, as I see it, those processes should be viewed as behaving like a swarm. Rather
than take living things apart, Artificial Life attempts to put living things together within a
bottom-up approach, that is, beyond life-as-we-know-it into the realm of life-as-it-could -be
(Langton), generating lifelike behaviour, and focusing on the problem of creating behaviour
generators, inspired on the nature itself, even if the results (what emerges from the process)
have no analogues in the natural world. The key insight into the natural method of behaviour
generation is gained by noting that nature is fundamentally parallel. This is reflected in the
“architecture” of natural living organisms, which consist of many millions of parts, each one
of which has its own behavioural repertoire. As we know, living systems are highly
distributed and quite massively parallel.
The so-called nouvelle Artificial Intelligence (AI) and aLife are each concerned with
the application of computers to the study of complex, natural phenomena. Apart from
traditional and symbolic hard-specific top-down AI in the sixties and seventies, both are
nowadays concerned with generating complex behaviour, in a bottom-up manner, turning
their attention from the mechanics of phenomena to the logic of it. The first computational
approach to the generation of lifelike behaviour was due to the mathematician John Von
Neumann. In the words of his colleague Arthur W. Burks, Von Neumann was interested in the
general question:
[…] What kind of logical organization is sufficient for an automaton to reproduce
itself ? This question is not precise and admits to trivial versions as well as interesting ones.
Von Neumann had the familiar natural phenomenon of self-reproduction in mind when he
posed it, but he was trying to simulate the self-reproduction of a natural system at the level of


4

genetics and biochemistry. He wished to abstract from the natural self-reproduction problem
its logical form […]
This approach is probably the first to capture the essence of Artificial Life (replace,
for instance, references to ’self-reproduction’ in the above with references to any other
biological phenomena). From this “kinematic model” of Von Neumann, a genuine selfreproduction mechanism implemented in the sixties, Stan Ulam suggested an appropriate
formalism where the logical form of the process is completely distinguish from the material
counterpart, which has come to be know as a Cellular Automata (CA). In brief, a CA consists
of a regular lattice of (many) finite automata , which are the simplest formal models of
machines. A finite automata can be in only one of a finite number of states at any given time,
and its transition between states from one time-step to the next are governed by a statetransition table : given a certain input and a certain internal state, the state-transition table
specifies the state to be adopted by the finite automata at the next time step. In a CA, the
necessary input is derived from the states of the automata at neighbouring lattice-points. Thus
the state of an cellular automata at time t+1 is a function of the states of the automata itself
and its immediate neighbours at time t. All the finite automata in the lattice (group of cells)
obey the same transition-table (rule table) and every cell changes his state at the same instant,
time-step after time-step. CA’s are a good example of the kind of computational paradigm
sought after by Artificial Life: bottom-up, parallel, local determination of behaviour with
minimal specification, and emerging complex phenomena from simple rules.
In order to study any natural phenomena, scientists are turning to a separation. A need
to separate the notion of a formal specification of a machine (any that will reproduce the
phenomena itself) – that is, a specification of the logical structure of the machine – from the
notion of a formal specification of a machines’s behaviour – that is, a specification of
transitions that the machine will undergo. In general, we cannot derive behaviours from
structure, nor can we derive structure from behaviours. So instead, in order to determine the
behaviour of some machines and coupled phenomena, there is no recourse but to run them
and see how they behave. This has consequences for the methods by which we (or nature) go
about generating behaviour generators themselves, and from which any evolutionary and
adaptive process seems to be essential. As an illustration, the most salient characteristic of
living systems, from the behaviour generation point of view, is the genotype/phenotype
distinction. The distinction is essentially one between a specification of machinery – the
genotype – and the behaviour of that machinery – the phenotype.
The genotype is the complete set of genetic instructions encoded in the linear
sequence of nucleotide bases that makes an organism’s DNA. The phenotype is the physical
organism itself – the structures that emerge in space and time as the result of the interpretation
of the genotype on a particular environment. The process by which the phenotype develops
through time under the direction of the genotype is called morphogenesis. Simulation plays an
essential role in the study of morphogenesis. This was anticipated as early as 1952 by Turing,
who wrote:
[…] The difficulties are such that one cannot hope to have any very embracing theory of such
processes, beyond the statement of equations. It might be possible, however, to treat a few
particular cases in detail with the aid of a digital computer. This method has the advantage
that it is not so necessary to make simplifying assumptions as it is when doing a more
theoretical type of analysis […]
What is notable is that these 1952 Turing words appears to have already t e embedded
h
features that characterise bottom-up approaches, in detriment of other kinds of approaches
strictly reductionist (e.g. top-down). As an aside evidence, note the last Turing words on this
sentence: it is not so necessary to make simplifying assumptions as it is when doing a more
theoretical type of analysis […]. Visualisation itself, of simulation results facilitates their
interpretation, and is used as a method for evaluating models. Lacking a formal measure of
what makes two patterns or forms (such as trees) look alike (task that is, as we known, mainly


5

related to the idea of perception), we rely on visual inspection comparing the models with the
reality. Important however in these models, is that the natural and synthetic pigmentation
patterns differ in details, yet we perceive them as fairly similar or familiar.
In morphogenesis, the individual genetic instructions are called genes and consist of
short stretches of DNA. These instructions are executed (expressed) when their DNA
sequence is used as a template for transcription. One may consider the genotype as a largely
unordered ‘bag’ of instructions (a rule table, an alphabet, a group of primitives), each one of
which is essentially the specification for a machine of some sort – passive or active. When
instantiated, each such machine will enter into ongoing logical mechanisms, consisting
largely of local interactions between other such machines. Each such instruction will be
executed when its own triggering conditions are met and will have specific, local effects on
structures in other cells (their neighbors). Furthermore, each such instruction will operate
within the context of all the other instructions that have been – or are being – executed.
The phenotype, then, consists of the structures and dynamics that emerge through
time in the course of the execution of the parallel, distributed computation controlled by this
genetic bag of instructions. Since genes interactions with one another are highly non-linear,
the phenotype is a non-linear function of the genotype. As mentioned briefly above, the
distinction between linear and non-linear systems is fundamental, and provides excellent
insight into why the principles underlying the dynamics of life (or many other natural
phenomena) should be so hard to find and understand. The simplest way to state the
distinction is to say that linear systems are those for which the behaviour of the whole is just
the sum of the behaviour of its parts, while for non-linear systems, the behaviour of the whole
is more than the sum of its parts. Linear systems are those which obey the principle of
superposition. We can break up complicated linear systems into simpler constituents parts,
and analyse these parts independently. Once we have reached an understanding of the parts in
isolation, we can achieve a full understanding of the whole system by composing our
understandings of the isolated parts. This is the key feature of linear systems: by studying the
parts in isolation we can learn everything we need to know about the complete system.
Nature, however, is generally non-linear, where this type of approach is often impossible.
Non-linear systems do not obey the principle of superposition. Even if we could break such
systems up into simpler constituents parts, and even if we could reach a complete
understanding of the parts in isolation, we would not be able to compose our understandings
of the individual parts into an understanding of the whole system. The key feature of nonlinear systems is that their primary behaviours of interest are properties of the interactions
between parts, rather than being properties of the parts themselves, and these interactionbased properties necessarily disappear when the parts are studied independently. Analysis has
not proved anywhere near as effective when applied to non-linear systems: the non-linear
system must be treated as a whole. A different approach to the study of non-linear systems
involves the inverse of analysis: synthesis. Rather than start with the behaviour of interest and
attempting to analyse it into its constituent parts, we should start with constituent parts and
put them together in the attempt to synthesize the behaviour of interest. Life, in the same way,
is a property of form, not matter, a result of organization and re-organization of matter rather
than something that inheres in the matter itself. Neither nucleotides nor amino acids nor any
other carbon-chain molecule is alive – yet put them together in the right way, and the dynamic
that emerges out of their interactions is what we call life. It is effects, not things, upon which
life is based – life is a kind of behaviour, not a kind of stuff – and as such, it is constituted of
simpler behaviours, not simpler stuff. Behaviours themselves can constitute the fundamental
parts of non-linear systems – virtual parts, which depend on non-linear interactions between
physical parts for their very existence. Isolate the physical parts and the virtual parts cease to
exist. It is the virtual parts of living systems that Artificial Life is after, and synthesis is its
primary methodological tool.
Computers, provide (and should be viewed as) as an important laboratory tool for the
study of life and many natural phenomena, as an alternative devoted exclusively to the
incubation of information structures. The advantage of working with information structures is
that information has no intrinsic size. The computer is the tool for the manipulation of


6

information, whether that manipulation is a consequence of our actions or a consequence of
the actions of the information structure themselves. Computers themselves will not be alive,
rather they will support informational universes within which dynamic populations of
informational ‘molecules’ (or memes, as proposed by Dawkins, as the cultural information
genes, or vehicles, within one specific society) engage in informational ‘biochemistry’. This
view of computers as workstations for performing scientific experiments within artificial
universes is fairly new, but is rapidly becoming accepted as a legitimate, even necessary, way
of pursuing science. In the days before computers, scientists worked primarily with systems
whose defining equations could be solved analytically, and ignored those whose defining
equations could not. This was the case, for instance, in many analytical systems trying to
explain how the global weather changes, or trying to forecast the behaviour of a fire
propagating in a specific terrain. As we now know, global weather is a chaotic non-linear
system, where a flap of a butterfly wing in Peking can develop a huge storm in New York,
few days later. In the absence of analytical possible solutions, the equations would have to be
integrated over and over again, essentially simulating the time behaviour of the system.
Without computers to handle the mundane details of these calculations, such an undertaking
was unthinkable except for the simplest cases. Given these mundane calculations to
computers, the realm of numerical simulation is opened up for exploration. ‘Exploration’ is an
appropriate term for the process, because the numerical simulation of systems allows one to
explore the system’s behaviour under a wide range of parameter settings and initial
conditions. The heuristic value of this experimentation cannot be overestimated. One often
gains tremendous insight for the essential dynamics of a system by observing its behaviour
under a wide range of initial conditions. Moreover, computers are beginning to provide
scientists with a new paradigm for modeling the world. When dealing with essentially
unsolvable governing equations, the primary reason for producing a formal mathematical
model (the hope of reaching an analytic solution by symbolic manipulation) is lost. It has
become possible, for example, to model turbulent flow in a fluid by simulating the motions of
its constituent particles – not just approximating changes in concentrations of particles at
particular points, but actually computing their motions exactly. The same is true for
understanding and modeling people in overcrowded soccer stadiums, or for instance, in
gaining insight on how traffic jams emerge, from very simple inner rules. Again, the best way
to tackle it, is to look at the whole process, synthesizing which basic and simple logical rules
(generally independent from the phenomena itself) govern the multitude of parts, emerging a
global and complex behaviour. What is essential in these types of systems, is not the parts and
their intrinsic natures (at least strictly), but namely their relationships (among themselves and
with their environment).
Within this same context, let us return again to the genotype/phenotype distinction
and on the possibility of the development of a behavioural phenotype. One paradigmatic
model is the one of Craig Reynolds, who in 1987 has implemented a simulation of flocking
behaviour. Now, if we think for a moment, none type of analytical differential equations was
been able to tackle (or model) this type of natural phenomena. In the Reynolds model,
however – which is meant to be a general platform for studying the qualitatively similar
phenomena of flocking, herding and schooling – one has a large collection of autonomous but
interacting objects (which Reynolds refer as Boids), inhabiting a common simulated
environment.
The modeler can specify the manner in which the individual Boids will respond to
local events or conditions. The global behaviour of the aggregate of Boids is strictly an
emergent phenomena, where none of the rules for the individual Boids depends on global
information, and the only updating of the global state is done on the basis of individual Boids
responding to local conditions. Note that, the underlying system nature is similar in many
ways to a Cellular Automata , mentioned earlier. Again, each Boid (cell for the CA) in the
aggregate shares the same behavioural ‘tendencies’:


7

·To maintain a minimum distance from other objects in the environment, including other
Boids,
·To match velocities with Boids in its neighbourhood, and
·To move towards the perceived centre of mass of the Boids in its neighbourhood.
These are the only rules governing the behaviour of the aggregate. These rules, then,
constitute the generalized genotype of the Boids system. What is amazing, is that they say
nothing about structure, or growth and development, or even about birds nature, but they
determine the behaviour of a set of interacting autonomous objects, out of which very natural
motion emerges. With the right settings for the parameters of the system, a collection of Boids
released at random positions within a volume will collect into a dynamic flock, which flies
around environmental obstacles in a very fluid and natural manner, occasionally breaking up
into sub-flocks as the flock flows around both sides of an obstacle. Once broken up into subflocks, the sub-flocks reorganize around their own, now distinct and isolated centre of mass,
only to re-emerge into a single flock again when both (or more) sub-flocks emerge at the fair
side of the obstacle and each sub-flock feels anew the mass of the other sub-flock.


8

The flocking behaviour itself constitutes the generalized phenotype of the Boids
system. It bears the same relation to the genotype as an organism’s morphological phenotype
bears to its molecular genotype. The same distinction, between the specification of machinery
and the behaviour of machinery is evident. Through development (or time), local rules
governing simple non-linear interactions at the lowest level of complexity emerge global
behaviours and structures at the highest level of complexity. Finally, Artificial Life (as a truly
interdisciplinary scientific field) may be viewed as an attempt to understand high-level


9

behaviour from low-level rules, for example, on how the simple interactions between ants and
their environment lead to complex trail-following behaviour. But by far more important than
studying ants itself, is to study how they organize themselves, through out a simple adaptive
mechanism that seems to be present in many natural phenomena of our world. An
understanding of such relationships in particular bio-inspired systems can suggest novel
solutions to complex real-world problems such as disease prevention, pattern recognition,
stock-market prediction, or data mining on the Internet (to name up a few).
One of the most well-know examples is the area of Evolutionary Computation. In the spirit of
Von Neumann, John Holland has attempted to abstract the logical form of the natural process
of biological evolution in what is currently known as the Genetic Algorithm (GA). In the GA,
a genotype is represented as a character string that encodes a potential solution to a problem.
For instance, the character string (chromosome) might encode the weight matrix of a neural
network, or the rule table of any Cellular Automata, or in its simplest way, any pseudosolution to any specific problem. These character strings are rendered as phenotypes via a
problem-specific interpreter, which constructs, for example, the artificial neural network or
the cellular automata machine specified by each genotype, evaluates its performance in the
problem domain, and provides it with a specific fitness value. From this point the GA
implements an artificial selection by making more copies of the character strings representing
the better performing phenotypes. The GA generates variant genotypes by applying genetic
operators to these character strings. The genetic operators typically consist of reproduction,
cross-over, and mutation, with occasional usage of inversion and duplication. What is
interesting is that “poor” individuals along several generations, often encode in parts of their
genotypes, the key for the best solutions (artificial individuals) to become better. The best GA
solution, is in some sense a product of the GA collective change of information, a product of
the whole, being diversity a key aspect in the process, and a way for the artificial algorithm to
balance his own exploration/exploitation duality character on the fitness landscape (space of
possible solutions). Such evolutionary approaches are being applied to tasks such as
optimisation, search procedures, classification, and adaptation, among others.
As the computational strategies mentioned above, Complex dynamic systems in
general show interesting and desirable behaviours as flexibility (in vision or speech
understanding tasks, the brain is able to cope with incorrect, ambiguous or distorted
information, or even to deal with unforeseen or new situations without showing abrupt
performance breakdown) or versatibility quoting Dorigo and Colorni, robustness (keep
functioning even when some parts are locally damaged - Damásio ), and they operate in a
massively parallel fashion. As we know, systems of this kind abound in nature. A vivid
example is provided by the behaviour of a society of termites (Courtouis). And, as a key
feature, complex dynamical systems show and provide emergent properties. Again, this means
that the behaviour of the system as a whole can no longer be viewed as a simple superposition
of the individual behaviours of its elements, but rather as a side effect of their collective
behaviour. Contained in this notion is the idea that properties are not a priory predictable
from the structure of the local interactions and that they are of functional significance. The
computation to be performed is contained in the dynamics of the system, which in turn is
determined by the nature of the local interactions between the many elements.
Many of the dynamical computation systems that have been developed today find
also their equivalent in nature, and all of them show, directly or not, important emergent
properties (among other lifelike features). A non-extensive list of possible paradigmatic
examples include, Genetic Algorithms, Memetic Algorithms, Spin Glass Models,
Connectionist Architectures and Artificial Neural Networks, Reaction-Diffusion systems, Self Organizing Maps, Simulated Annealing methods, Artificial Imunne systems, Cellular
Automata , L-Systems, Gradient Vector Flow and Snakes, Differential Evolution,
Correlational Opponent Processing and Particle Swarm Optimization.


10

As an example, biological metaphors offer insight into many aspects of computer
viruses and can inspire defences against them. That is the case with some applications of
Immunological Computation, and Artificial Immune Systems. The immune system is highly
distributed, highly adaptive, maintains a memory of past encounters, is self-organising in his
own nature and has the ability to continually learn about new encounters. From a
computational viewpoint, the immune system has much to offer by way of inspiration.
Detection of specific patterns in large databases is one possible application. Autonomous alert
collision systems, in route management for airplanes is another.
An important feature in many of these dynamical computational systems is that of
interaction (e.g. competition-cooperation duality). Cooperation involves a collection of agents
– global behaviours, if we strictly follow Langton words - that interact by communicating
information, or hints (usually concerning regions to avoid or likely to contain solutions) to
each other while solving a problem. This duality interaction can also be found in the wellknown Prisoner Dilemma game theory problem, into which many Evolutionary Computation
approaches are being used. The information exchanged may be incorrect at times and should
alter the behaviour of the agents receiving it, yet, what emerges at the end is an robust rule in
the pool of rules, which is cooperative. Another example of cooperative problem solving is
the use of the Genetic Algorithm to find states of high fitness in some abstract space. In a
Genetic Algorithm, members of a population of states exchange pieces of themselves or


11

mutate to create new populations, often containing states of higher fitness. In Artificial Neural
Networks, we can also find similar features, where the output of one neuron affects the
behaviour (or state – under the light of Cellular Automata theory) of the neuron receiving it,
and so on. Reporting to the real nature and quoting Damásio, we are barely beginning to
address the fact that interactions among many non-contiguous brain regions probably yield
highly complex biological states that are vastly more than the sum of their parts. It is
important, however, t point out that the brain and mind are not a monolith: they have
o
multiple structural levels, and the highest of those levels creates instruments or artefacts that
allow for the observation of the other levels.

MOVING ON TO THE IMPLICIT
Evolutionary Computation, whether in the form of Genetic algorithms, Genetic
Programming, or Evolution Strategies, has been largely successful in solving a great variety
of problems in many scientific areas. This is in part due to the fact that for these types of
applications, an appropriate fitness evaluation is possible to code, and depending on the
problem is a relatively simple task, that is, a performance of each solution in the population
can be measured. Even if Artificial Life has made a strong rupture with the more traditional
symbolic AI, by implementing non-analytical bottom-up approaches, yet and for some
algorithmic paradigms like Evolutionary Computation, there is still a need for a high-level
specification of purpose (or intention), a target, in order to evaluate and select solutions found
so far in each generation. However, in some real-world implementations where these
evaluations are hard to formalize by any group of equations, being it in the form of multiple
coded constraints, specific grammars, confidence intervals or as normally by any multiobjective evaluation function, the successful and coherent application of Genetic Algorithms
are jeopardize, remaining the strategy a pure random process. This is the case, in many, if not
all the recent implementations of synthetic evolutionary art, or generative art and architecture.
As we know, defining any aesthetic criteria is difficult, being the translation from these to an
automatic set of mathematical selection rules, probably even more difficult or impossible
(since, among other aspects, the relation of the art work in formation and the artist can be seen
as a process of co-evolution). In the absence of any mathematical function that can map
coherently the relations of, for instance, form into aesthetical value, the final result will
always be a random guided-tour of some sub-space of possible and hypothetical novel
solutions, not different in many aspects, to a trial and error basic process, submitted to any
specific conceptual search space. This is mainly a problem of representation, since any
attitude to implement those aesthetical fitness functions, mapping genotypes into the
“usefulness” of the hypothetical novel forms, can be as dangerous as the objective and
analytical evaluation of any final art work. This high-level mapping attitude is in itself a
compression method, where the diversity of any conceptual world and the nature of its several
dimensions are reduced to some aspects. Luis Borges words on these matters are wise: the
only true map of the world, is the world itself.
There is however one way to avoid those mathematical mappings. That is of using the
human observer or “artist”, as an “aesthetical mapping machine”, connected in real-time to
the artificial evolutionary process. The first computer-graphics program where the idea was
introduced was due to Karl Sims (1991). This program uses genetic algorithms to generate
new images, or patterns, from pre-existing images. Unlike most GA systems, the selection of
the “fittest” examples is not automatic, but is done by the programmer. That is, the human
being selects the images which are aesthetically pleasing, or otherwise interesting, and these
are used to breed the next generation, being the whole an interactive graphics environment.
Although interesting in many fields, this type of Evolutionary Computation is however
dependent on the human observer, and on his attitude, since the measuring of any pool of
pseudo-solutions (e.g., art works in formation), at any given generation, must be evaluated
and ranked by the artist looking to the monitor. For instance, there are several Genetic
Algorithms, that work under this line, helping to find criminals. In order to search optimally


12

all the combinatorial nature of the human faces, the victim points out to the GA, some aspects
and features of the face of the criminal, evaluating and ranking different evolutionary
proposals, generation after generation, being the final result a robot-drawing of that criminal
(or at least, similar to him). In the context of artificia l art, this conceptual framework has
recently being followed, but still, the human-computer interaction, determines a final art
work, which is far from being purely a result from an autonomous process, being the
computational paradigm just a tool to achieve any artistic purpose, as a pencil still his for any
drawing artist.
Even if those methodological approaches are interesting, there is still a profound gap
on the understanding of other mechanisms that can have a crucial role on the sciences of the
artif icial, and predominantly on the nature of morphogenesis. As an example, for the past ten
years or so, aLife research debates intensively the basic characters of those non-linear
mappings between genotypes and phenotypes. What are, for instance, the necessary alphabets
and their local relationships, which can promote and emerge spatial organization at the higher
levels of complexity in any system? What parts and elementary mechanisms are essential, in
order to have the system behaving as a usefulness, creative and autonomous whole? In order
to find a truly innate artificial emergent life (a research field that could have an enormous
impact on the synthetic computational art and architecture, itself), we must thus, study and
find other approaches, which can be coupled or not with the existing aLife systems.
As a researcher in all the computational paradigms briefly described above, I believe
that there are three crucial aspects in order to create truly low-level mechanisms that could
autonomously emerge novel aesthetical patterns. The first aspect to discuss, is on the nature
of autonomy itself, that is, on the nature of any autonomous mechanisms embedded in any
artificial organism. Rather specifying high-level constraints, even if minimally at each
chromosome fitness evaluation, we should take the opposite way, that is, to allow for any
organism an implicit nature of those emerged characters. We must follow the bottom-up
methodological design, till the end. The second is on the role of intrinsic co-evolution
between parts of the artificial system, on how it can be implemented, and on his intrinsic
scientific properties. By doing this, we allow the system to connect into any and possible
unsupervised realm. However, and for the sake of simplicity, I will not discuss these
properties on here. Finally the third aspect is related to the nature of self-organization, a
concept that links back to autonomy and emergence.
The first aspect of autonomy involves behaviour mediated, in part, by inner
mechanisms shaped by any artificial organism past experience. These mechanisms may, but
need not, include explicit representations of current or future states. However, the important
distinction is between a response wholly dependent on the current environmental state (given
the or iginal, “innate”, bodily mechanisms), and one largely influenced by the creature’s
experience. The more a creature’s past experience differs from that of other creatures, the
more “individual” its behaviour will appear. The second aspect of autonomy, relates to know
into what extend the controlling mechanisms were self-generated rather than externally
imposed. That is, a distinction between behaviour which emerges as a result of selforganizing processes, and behaviour which was deliberately prefigured in the design of that
organism, or organisms. This concerns for instance those behaviours that emerge, from an
initial list of simple rules concerning locally interacting units, but it was neither specifically
mentioned in those rules, nor foreseen when they were written. As an example, this is case of
Boids. Ethnologists, aLife workers, and situated roboticists (e.g., Inman Harvey), all assume
that increasingly complex hierarchical behaviour can arise in this sort of way. The more levels
in the hierarchy (layers of complexity, as I like to call them), the less direct the influence of
environmental stimuli will be – and the greater the behavioural autonomy. Even if, all started
(or can be started, as I believe), from a simple set of environmental stimuli. A primordial soup
of implicit characters. An intrinsic and kaleidoscopically genotype.
For all these reasons and within all these paradigms, we must focus our study and
implement computational features that underlie the collective and the distributed, the flexib le
and the versatile, the massively parallel and the dynamical. Finally we should research those
systems that rely on synergy, cooperation and co-evolution, with or without embodied


13

evolutionary computation. One prominent example, are Artificial Ant Systems, a research area
to which I humble contribute since 1998. In “
Godel, Escher, Bach”, Douglas Hofstadter
explores the difference between an ant colony as a whole and the individual that composes it.
According to Hofstadter, the behaviour of the whole colony is far more sophisticated and of
very different character than the behaviour of the individual ants. A colony’s collective
behaviour exceeds the sum of its individual member’s actions (so-called emergence) and is
most easily observed when studying their foraging activity. Most species of ants forage
collectively using chemical recruitment strategies, designated by pheromone trails, to lead
their fellow nest-mates to food sources.
This analogy with the way that real and natural ant colonies work and migrate, has
suggested the definition in 1991/92 of a new computational paradigm, which is called the Ant
System (Dorigo / Colorni). In these studies (again) there is no pre-commitment to any
particular representational scheme: the desired behaviour is specifie d, but there is minimal
specification of the mechanism required to generate that behaviour, i.e. global behaviour
evolves from the many relations of multiple simple behaviours. Since then several studies
were conducted to apply this recent paradigm – or analogous ones - in real case problems,
with successful results. In short, the new heuristic has the following desirable characteristics:
(1) It is versatile , in that it can be applied to similar versions of the same problem; (2) It is
Robust. It can be applied with only minimal changes to other problems (e.g. combinatorial
optimisation problems such as the quadratic assignment problem - QAP, travelling salesman
problem - TSP, or the job-shop scheduling problem - JSP);…and (3) It is a population based
approach. This last property is interesting since it allows the exploitation of positive feedback
as a search mechanism (the collective behaviour that emerges is a form of autocatalytic
“snow ball” - that reinforces itself - behaviour, where the more the ants follow a trail, the
more attractive that trail becomes for being followed). It also makes the system amenable to
parallel implementations (though, only the intrinsically parallel and distributed nature of these
systems are generally considered).
One typical case of interest is that of perception. I have explored the application of
Artificial Ant Systems into Pattern Recognition problems, namely to the sub-problem of
image segmentation, i.e., to find homogeneous regions in any digital image, in order to extract
and classify them. The application of these heuristics onto image segmentation looks very
promising, since segmentation can be looked as a clustering and combinatorial problem, and
the grey level image itself as a topographic map (where the image is the ant colony
playground). Then, the distribution of the pheromone (a volatile and chemical substance)
represents the memory of the recent history of the swarm, and in a sense it contain
information which the individual ants are unable to hold or transmit. In this artificial system,
there is no direct communication between the organisms but a type of indirect communication
through the pheromonal field. In fact, ants are not allowed to have any memory and the
individual’s spatial knowledge is restricted to local information about the whole colony
pheromone density. Particularly interesting for the present task (i.e. trying to evolve
perceptive capabilities), is the fact that self-organisation of ants into a swarm and the selforganisation of neurones into a brain-like structure are similar in many respects (Chialvo,
Millonas). Swarms of social insects construct trails and networks of regular traffic via a
process of pheromone laying and following. These patterns constitute what is known in brain
science as a cognitive map. The main differences lies in the fact that insects write their spatial
memories in the environment, while the mammalian cognitive map lies inside the brain, a fact
that also constitutes an important advantage in the present model. As mentioned by Chialvo,
this analogy can be more than a poetic image, and can be further justified by a direct
comparison with the neural processes associated with the construction of cognitive maps in
the hippocampus. Wilson, for instance, forecasted the eventual appearance of what he called
“a stochastic theory of mass behaviour” and asserted that “the reconstruction of mass
behaviours from the behaviours of single colony members is the central problem of insect
socio-biology”. He forecasted that our understanding of i dividual insect behaviour together
n
with the sophistication with which we will able to analyse their collective interaction would
advance to the point were we would one day posses a detailed, even quantitative,


14

understanding of how individual “probability matrices” would lead to mass action at the level
of the colony. By replacing colony members with neurones, mass behaviours or colony by
brain behaviour, and insect socio-biology with brain science the above paragraph could
describe the paradigm shifts in the last twenty-five years of progress in the brain sciences.
Perception is also an important conceptual element in what can be related to
autonomy. Probably a third criterion of autonomy (not listed earlier) links to the extend to
which a system’s inner directing mechanisms can be reflected upon, and/or selectively
modified, by the individual concerned. One way in which a system can adapt its own
processes, selecting the most fruitful modifications, is to use an evolutionary strategy such as
the genetic algorithms mentioned above. It may be that something broadly similar goes on in
human minds. But the mutations and selections carried out by GAs are modelled on biological
evolution, not conscious reflection and self-modification. And it is conscious deliberation
which many people assume to be the root of human autonomy. Thus, it is primarily on
perception and inner recognition that the system must rely.
Moreover, perception itself, as a human feature is being modelled and analysed by
Gestalt psychology and philosophical systems since, at least 1910 (Wertheimer). It is of much
interest to follow that this kind of scientific works point out that perception is a product of a
synergistic whole effect, i.e. the effect of perception is generated not so much by its
individual elements (e.g. human neurones) as by their dynamic interrelation (collective
behaviour) – phenomena that can be found easily in many computational paradigms briefly
described above, or even in Neural Network computational models, where data generalisation,
N dimensional matrix re-mapping, pattern classification or forecasting abilities are known to
be possible. As putted by Limin Fu in his own words, the intelligence of a Neural Network
emerges from the collective behaviour of neurones, each of which performs only very limited
operations. Even though each individual neuron works slowly, they can still quickly find a
solution by working in parallel. This fact can explain why humans can recognize a visual
scene faster than a digital computer, while an individual brain cell responds much more
slowly than a digital cell in a VLSI (Very Large Scale Integration) circuit. Also, this brain
metaphor suggests how to build an intelligent system which can tolerate faults (fault
tolerance) by distributing information redundantly. It would be easier to build a large system
in which most of the components work correctly than to build a smaller system in which all
components are perfect. Another feature exhibited by the brain is the associative type of
memory. The brain naturally associates one thing with another. It can access information
based on contents rather than on sequential addresses as in the normal digital computer. The
associative, or content-addressable, memory accounts for fast information retrieval and
permits partial or approximate matching. The brain seems to be good at managing fuzzy
information because of the way its knowledge is represented. The key aspect is that artificial
ant systems behave similarly.
Typically these systems form a structure, configuration, or pattern of physical,
biological, sociological, or psychological phenomena, so integrated as to constitute a
functional unit with properties not derivable from its parts in summation (i.e. non-linear) –
Gestalt in one word (Krippendorff) (the English word more similar is perhaps system,
configuration or whole ). This synergetic view, derives from the holistic conviction that the
whole is more than the sum of its parts and, since the energy in a whole cannot exceed the
sum of the energies invested in each of its parts (e.g. first law of thermodynamics), that there
must therefore be some quantity with respect to which the whole differs from the mere
aggregate. This quantity is called synergy and in many aLife computational systems can be
seen as their inherent emergent and autocatalytic properties (process well known in many
Reinforcement Learning models, namely in Q-learning methods often used in autonomousagents design (Mitchell / Maes).
Part of what we now see in these figures, was due to a model that has explored the
application of these features into digital images, replacing the normal colony habitat, by grey
levels, extending the capabilities of pheromone deposition into different situations, allowing a
process of perceptual morphogenesis. In other words, from local and simple interactions to
global and flexible adaptive perception. In those experiments, the emergence of network


15

pheromone trails, for instance, are the product of several simple and local interactions that
can evolve to complex patterns, which in some sense translate a meta -behaviour of that
swarm. Moreover, the translation of one kind of low-level structure of information (present in
a large number) to one meta-level is minimal. Although that behaviour is specified (and
somehow constrained), there is minimal specification of the mechanism required to generate
that behaviour; global behaviour evolves from the many relations of multiple simple
behaviours, without global coordination, and using indirect communication (through the
environment). One abstract example is the notion, within a specified population, of commonsense, being the meta-result a type of collective-conscience. Needless to say, that some
features are acquired (through out the evolving relation with the habitat), being others inner
components of each part. Though, what is interesting to note is that we do not need to specify
them. Moreover, the present model shows important adaptive capabilities, as in the presence
of sudden changes in the habitat. Even if the model is able to quickly adapts to one specific
environment, evolving from one empty pheromonal field, habitat transitions point that, the
whole system is able to have some memory from past environments (i.e. convergence is more
difficult after learning and perceiving one habitat). This emerged feature of résistance, is
somewhat present in many of the natural phenomena that we find today in our society.
In short, the design of such systems must follow a conceptual flux, where autonomy,
perception and synergy are the key-elements. My final words are exactly about synergy
within ant systems, and on how this aLife scientific essay in the intersection can help building
or suggest novel 2D patterns, or even 3D architectures, as we now see on these pages.
Synergy (from the Greek word synergos), broadly defined, refers to combined or co-operative
effects produced by two or more elements (parts or individuals). The definition is often
associated with the quote “the whole is greater than the sum of its parts” (Aristotle , in
Metaphysics), even if it is more accurate to say that the functional effects produced by wholes
are different from what the parts can produce alone. Synergy is a ubiquitous phenomenon in
nature and human societies alike. One well know example is provided by the emergence of
self-organization in social insects, via direct (mandibular, antennation, chemical or visual
contact, etc) or indirect interactions. The latter types are more subtle and defined by Grassé as
stigmergy to explain task coordination and regulation in the context of nest reconstruction in
Macrotermes termites. An example, could be provided by two individuals, who interact
indirectly when one of them modifies the environment and the other responds to the new
environment at a later time. In other words, stigmergy could be defined as a typical case of
environmental synergy. Grassé showed that the coordination and regulation of building
activities do not depend on the workers themselves but are mainly achieved by the nest
structure: a stimula ting configuration triggers the response of a termite worker, transforming
the configuration into another configuration that may trigger in turn another (possibly
different) action performed by the same termite or any other worker in the colony. Another
illustration of how stigmergy and self-organization can be combined into more subtle adaptive
behaviors is recruitment in social insects. Self-organized trail laying by individual ants is a
way of modifying the environment to communicate with nest mates that follow such trails.
It appears that task performance by some workers decreases the need for more task
performance: for instance, nest cleaning by some workers reduces the need for nest cleaning.
Therefore, nest mates communicate to other nest mates by modifying the environment
(cleaning the nest), and nest mates respond to the modified environment (by not engaging in
nest cleaning); that is stigmergy.
In other words, perception and action only by themselves can evolve adaptive and flexible
problem-solving mechanisms, or emerge communication among many parts. The whole and
their relationships (that is, the next layer in complexity) emerges from the relationship of
many parts, even if these latter are acting strictly within and according to any sub-level of
basic and simple strategies, ad-infinitum repeated. Quoting Einstein , the system “should be
made as simple as possible, but not simpler”.


16

On the rigth, a sequential clustering task of corpses performed by a real ant colony. 1500
corpses are randomly located in a circular arena with radius = 25 cm, where Messor Sancta
workers are present. The figure shows the initial state (a), 2 hours (b), 6 hours (c) and 26
hours (d) after the beginning of the experiment. On the left, an artificial swarm evolving
clusters of semantically similar data items (Ramos, 2001).

Division of labor is another paradigmatic phenomena of stigmergy. Simultaneous task
performance (parallelism) by specialized workers is believed to be more efficient than
sequentia l task performance by unspecialized workers. Parallelism avoids task switching,
which costs energy and time. A key feature of division of labor is its plasticity. Division of
labor is rarely rigid. The ratios of workers performing the different tasks that maintain the
colony’s viability and reproductive success can vary in response to internal perturbations or
external challenges.
But by far more crucial to the design of any collective pattern artificial system, is how
ants form piles of items such as dead bodies (corpses), larvae, or grains of sand. There again,
stigmergy is at work: ants deposit items at initially random locations. When other ants
perceive deposited items, they are stimulated to deposit items next to them, being this type of
cemetery clustering organization and brood sorting a type of self-organization and adaptive
behavior. Théraulaz and Bonabeau described for instance, a model of nest building in wasps,
in which wasp-like agents are stimulated to deposit bricks when they encounter specific
configurations of bricks: depositing a brick modifies the environment and hence the
stimulatory field of other agents. These asynchronous automata (designed by an ensemble of


17

A swarm emerging global 3D patterns, from multiple local configuration stimulus.

algorithms) move in a 3D discrete space and behave locally in space and time on a pure
stimulus-response basis. There are other types of examples (e.g. prey collectively transport),
yet stimergy is also present: ants change the perceived environment of other ants (their
cognitive map, according to Chialvo and Millonas), and in every example, the environment
serves as medium of communication.
What all these examples have in common is that they show how stigmergy can easily
be made operational. As mentioned by Bonabeau, that is a promising first step to design
groups of artificial agents which solve problems: replacing coordination (and possible some
hierarchy) through direct communications by indirect interactions is appealing if one wishes
to design simple agents and reduce communication among agents. Another feature shared by
several of the examples is incremental construction: for instance, termites make use of what
other termites have constructed to contribute their own piece. In the context of optimization
(though not used on the present works), incremental improvement is widely used: a new
solution is constructed from previous solutions (see ACO paradigm, Dorigo et al). Finally,
stigmergy is often associated with flexibility: when the environment changes because of an
external perturbation, the insects respond appropriately to that perturbation, as if it were a
modification of the environment caused by the colony’s activities. In other words, the colony
can collectively respond to the perturbation with individuals exhibiting the same behavior.
When it comes to artificial agents, this type of flexibility is priceless: it means that the agents
can respond to a perturbation without being reprogrammed in its intrinsic features to deal with
that particular instability. The system organizes itself in order to deal with new object classes
(conceptual ideas translated to the computer in the form of basic 2D/3D forms), or even new
sub-classes. This task can be performed in real time, and in robust ways due to system’s
redundancy.
Data and information clustering is one of those problems in which real ants can
suggest very interesting heuristics for computer scientists, and it is in fact a classic strategy
often used in Image and Signal Processing. For the past two years, I have been developing
research on these areas. Many experiments are now under their way at the CVRM-IST Lab
(for instance, real-time marble and granite image classification, image and data retrieval, etc),
along with the application of Genetic Algorithms, Neural Networks, and many others (based
strictly on natural computation paradigms) into many problems in Natural Resources
Management, like forecasting water quality and control on river networks. Surprisingly, these
studies can help us to understand how artificial stigmergic systems can be implemented in
order to produce novel and autonomous patterns.


18

Jackson Pollock

These algorithms mimic the inner stimuli response threshold functions of each
organism composing the system, and in some sort, what happens in several species of ants,
where workers have been reported to sort their larvae or form piles of corpses – literally
cemeteries – to clean up their nests. Chrétien has performed experiments with the ant Lasius
niger to study the organization of cemeteries. Other experiments include the ants Pheidole
pallidula reported by Denebourg, and many species actually organize a cemetery. If corpses,
or more precisely, sufficiently large parts of corpses are randomly distributed in space at the
beginning of the experiment, the workers form cemetery clusters within a few hours,
following a behavior similar to aggregation. If the experimental arena is not sufficiently large,
or if it contains spatial heterogeneities, the clusters will be formed al ng the edges of the
o
arena or, more generally, following the heterogeneities. The basic mechanism underlying this
type of aggregation phenomenon is an attraction between dead items mediated by the ant
workers: small clusters of items grow by attracting workers to deposit more items. It is this
positive and autocatalytic feedback that leads to the formation of larger an larger clusters. In
this case, it is therefore the distribution of the clusters in the environment that plays the role of
stigmergic variable.
Finally, the simulated ecology of different stimuli response threshold organisms,
triggered by the seeds of these stigmergic processes, whether in the form of 3D local
configurations, or by the qualitative values of any conceptual data items, must not be
overestimated. Above all, the behaviour that emerges from all these spatial temporal
relationships conduct us into the realm of what nature is about: dynamical patterns of
complexity. Not chaotic or purely rendered at random, but at the edge of chaos (Langton),
where creative and autonomous aLife survives. As reported recently by Nature magazine
(Sept., 13, 2000), research suggests that the abstract works of artists such as Jackson Pollock
are esthetically pleasing because they obey fractal rules similar to those found on the natural
world. Pollock was known to have swung his paint back and forth like a pendulum, using a
can on the end of a string with a hole punched in it. Researchers (Jensen) have found that if a
swinging pendulum is hit with a hammer at just the right frequency (slightly less than the
natural rhythm of the pendulum), its motion becomes chaotic and the paint traces out very
convincing “fake Pollocks”. However, the artist had no idea of fractals or chaotic motion.
This seems to be in line with the actual synthetically computational art, where there is a need
to reference some kind of external artifact or mechanism, but nevertheless and as it appears,
not those of the self whether they are conscious, unconscious, intuitive or not. Synthetically
generative art, and above all, artificial systems of morphogenesis of any kind, should be much
more about what scientists call “complexity”, and rely on nature as a physical generative
force of ontological significance. Moving on to the implicit, rather on the specific.


19

FURTHER READING MATERIAL:
[http://www.lxxl.pt/aswarm/aswarm.html]:
Leonel Moura, Swarm Paintings – Non-Human Art, in ARCHITOPIA Book / Catalogue, Art, Architecture and
Science, J.L. Maubant and L. Moura (Eds.), pp. 1-24, Ministério da Ciência e Tecnologia, Feb. 2002.

[http://alfa.ist.utl.pt/~cvrm/staff/vramos]:
[ ] Vitorino Ramos, Filipe Almeida, Artificial Ant Colonies in Digital Image Habitats - A Mass Behaviour
Effect Study on Pattern Recognition, ANTS'2000 – 2nd Int. Workshop on Ant Algorithms (From Ant Colonies to
Artificial Ants), M. Dorigo, M. Middendorf & T. Stüzle (Eds.), Brussels, Belgium, 7-9 Sep. 2000.
[ ] Vitorino Ramos, The Biological Concept of Neoteny in Evolutionary Colour Image Segmentation - Simple
Experiments in Simple Non-Memetic Genetic Algorithms, in Applications of Evolutionary Computation,
E.J.W. Boers et al. (Eds.), SPRINGER-VERLAG, 2001.
[ ] Vitorino Ramos, Fernando Muge, Map Segmentation by Colour Cube Genetic K-Mean Clustering, in
Research and Advanced Technology for Digital Libraries, J. Borbinha and T. Baker (Eds.), SPRINGERVERLAG, 2000.
[ ] Vitorino Ramos, Self-Organized Data and Image Retrieval as a Consequence of Inter-Dynamic
Synergistic Relationships in Artificial Ant Colonies, ICEIS´02 – 4th Int. Conf. on Enterprise Information
Systems, Ciudad-Real, Spain, 3-6 April, 2002.
[ ] Vitorino Ramos, Juan J. Merelo, Self-Organized Stigmergic Document Maps: Environment as a
Mechanism for Context Learning, AEB´2002 – 1st Spanish Conference on Evolutionary and Bio-Inspired
Algorithms, Mérida, Spain, 6-8 Feb. 2002.
[ ] Vitorino Ramos, Leonel Moura, The MC2 Project [Machines of Collective Conscience], in the Official
Newspaper of the UTOPIA Biennial Art Exposition, Cascais, Portugal, July 12-22, 2001.
[ ] Vitorino Ramos, Luis Ribeiro, Exploring Temporal Rule-Relations in Water Quality Time Series with
Swarm Intelligence , AIS´2002 - Artificial Intelligence, Simulation and Planning in High Autonomy Systems,
Lisbon, Portugal, April 7-10, 2002.
[ ] Vitorino Ramos, Pedro Pina, Fernando Muge, Mining Textural Features with Artificial Ant Colonies –
Towards a Dynamic and Continuous Classification of Polished Marble Slabs, RecPad’2002, 12th Int.
Portuguese Conf. On Pattern Recognition, Aveiro, Portugal, 2002.
[ ] Vitorino Ramos, Stigmergic Design: Collective Patterns in 3D-Space and Time, DESIGN & NATURE
2002, WIT Press, Udine, Italy, 2002.
[ ] Vitorino Ramos, From Biological Models of Stigmergy in Social Insects to Collective, Dynamical and SelfOrganized Data Mining Systems, DATA MINING 2002, WIT Press, Bologna, Italy, 2002.

Vitorino Ramos: on the implicit and on the artificial

More Related Content

What's hot (19)

Viewers also liked (17)

Similar to Vitorino Ramos: on the implicit and on the artificial (20)

More from ArchiLab 7 (20)

Recently uploaded (20)

Vitorino Ramos: on the implicit and on the artificial