2005: natural computing - concepts and applications

Natural Computing:Natural Computing:A Brief Survey of Ideas and A Brief Survey of Ideas and

ApplicationsApplications

BIC 2005: BIC 2005: International Symposium on Bio-Inspired ComputingInternational Symposium on Bio-Inspired Computing

Johor, MY, 9Johor, MY, 9thth September 2005 September 2005

Dr. Leandro Nunes de [email protected]

http://lsin.unisantos.b/lnunesCatholic University of Santos - UniSantos/Brazil

BIC 2005 - Natural Computing - Dr. Leandro Nunes de Castro 2

Imagine a world where computers can create new universes, and within these universes there are natural forms that reproduce, grow and adapt. Imagine natural patterns, mountains, ant colonies, immune systems and brains, all learning and evolving, and becoming increasingly more adapted to the environment. Imagine if our computers could contain new forms of life. Think how this would affect our lives. Maybe we could automatically create house and music design, new forms of protecting computers against invaders, new forms of solving complex problems, new organisms and new forms of computing.

Now stop imagining.Welcome to Computing in the New Millennium. Welcome to the Natural Computing age!

Foreword

Adapted from Digital Biology, by P. Bentley.


Outline

Part I: Introduction and Motivation Some ideas and challenges

Part II: Looking at Nature with Different Eyes Nature’s solutions: Some samples

Part III: Natural Computing Computing inspired by nature The simulation and emulation of natural

phenomena in computers Computing with natural materials

Part III: Computing in the New Millennium

Part I

Introduction and MotivationIntroduction and Motivation


Current Computer Technology

Turing Machines (TM) Computational device idealized by A. Turing in

1936 If a problem can be computed, then it can be

computed by a Turing Machine

J. von Neumman architecture


Features of Current Computers

General-purpose machines Manipulate precisely precise information* Address-based memory Serial processing* Are not capable of generalizing Are not fault tolerant (robust) Are not adaptable* …


Are You Ready?

1. Develop a computer program to distribute products of a company throughout the country.

2. Generate a computer model to simulate the evacuation program of a building undergoing fire.

3. What are the new technologies to complement or supplement silicon-based hardware?


Why Are These Questions Hard?(1. Products Distribution)

How many are the possible routes?


Why Are These Questions Hard?(2. Behavioral Simulation)


Why Are These Questions Hard?

(3. New Technologies)

Moore’s Law: The processing power

of silicon-based computers doubles approximately every couple of years

By the end of the next decade (2020) we may have reached the (miniaturization) limit of current computer technology

N.

of

ato

ms

per

bit

Year

2020: 1 atom per bit


What all these questions have in common?

The answer to all of them require a paradigm shift

Where can we find answers to them?Where can we find answers to them?

• Where all these problems and difficulties have been solved and dealt with from ages: In NATURE!!

Part II

Looking at Nature with Different Eyes----

Nature’s Solutions: Some Samples


Natural Architects


Natural Deliverers and Cleaners


Natural Behavior Animators


Natural Computer

Part III

Natural ComputingNatural Computing


From Nature to Computing: Natural Computing

Nature x Computing Natural computing is the terminology

used to encompass three paradigms: Computing inspired by nature The simulation and emulation of natural

phenomena in computers Computing with natural materials


The Philosophy of Natural Computing

Part III-A

Computing Inspired by NatureComputing Inspired by Nature


Main Ideas

Nature has evolved through ages in order to solve complex real-world problems

Examples abound: nest building, nest cleaning, main senses (hearing, seeing, touching, smelling, tasting), etc.

Computer algorithms based or inspired by nature have been developed for some time: Either to model nature, Or to solve complex real-world problems


Main Themes

Neurocomputing Evolutionary Computing Swarm Intelligence Immunocomputing Artificial Chemistry Growth and Developmental Algorithms etc.

Older approaches


Neurocomputing

Inspiration


Design principles: Artificial neuron: basic information

processing and storage unit Network architecture: how the artificial

neurons are interconnected Learning algorithm: guides the dynamics

(adaptability) of the system

Neurocomputing


Neurocomputing

Basic artificial neuron Some activation

functions


Network architectures Single-layer feedforward network

Neurocomputing


Network architectures Multi-layer feedforward network

Neurocomputing


Network architectures Recurrent network

Neurocomputing


Learning algorithms/rules: Hebb learning Single-layer perceptron Adaline ART Multi-Layer perceptron Self-organizing networks Hopfield networks Grossberg networks …

Neurocomputing


Why neurocomputing? Learning capability Parallel processing Generalization capability Inherently distributed Robust ...

Neurocomputing


Scope: Function approximation Clustering Classification Pattern recognition Control …

Mature field with innumerable academic, industrial, commercial and governmental applications

Neurocomputing


+ Reproduction

+ Genetic Variation

+ Selection

Evolutionary Computing

Inspiration


The power of (artificial) evolution




The power of evolution


Design principles: Population of individuals* Reproduction with genetic inheritance Genetic variation Selection



Standard evolutionary algorithm


procedure [P] = standard_EA(pc,pm)

initialize Pf eval(P)P select(P,f)t 1while not_stopping_criterion

do,P reproduce(P,f,pc)P variate(P,pm)f eval(P)P select(P,f)t t + 1

end whileend procedure


Main types of evolutionary algorithms: Evolutionary programming Evolution strategies Genetic algorithms Genetic programming* Classifier systems*



Why evolutionary computing? A population may explore and exploit more

efficiently than a single individual Importance of information (experience)

exchange Maintenance of good quality solutions Diversity and creativity



Scope: Search and optimization Planning (e.g. routing, scheduling and packing

) Design (e.g. signal processing) Simulation, identification, control (e.g. general

plant control) Classification (e.g. machine learning, pattern

recognition and classification)



Systems based on the collective behavior of social organisms

Two main approaches: Systems based on the collective behavior of

social insects Ant Colony Optimization (ACO) Ant Clustering Algorithm (ACA)

Systems based on sociocognition Particle Swarm Optimization (PSO)

Swarm Intelligence


An inspiration

Swarm Intelligence


Swarm Intelligence An ant farm


An ant farm

Swarm Intelligence


Another inspiration

Swarm Intelligence


Robotic autonomous navigation

Swarm Intelligence


Why swarm intelligence? Again, a multi-agent approach may allow for a

better exploration and exploitation of the space

Simple agents together can perform complicated tasks

It may be easier and cheaper to have many simple agents than a single complex one

Swarm Intelligence


Scope: Search and optimization:

Discrete and continuous optimization Data analysis (clustering) Robotics (autonomous navigation)

Swarm Intelligence


Immunocomputing

Inspiration


Design principles: Representation Architecture Affinity/Fitness functions Dynamics/Metadynamics

Immunocomputing


Representation Set of coordinates: m = m1, m2, ..., mL, m

SL L Ab = Ab1, Ab2, ..., AbL,

Ag = Ag1, Ag2, ..., AgL Some Types of Shape Space

Hamming Euclidean Manhattan Symbolic

Immunocomputing


Immunocomputing

Affinities: related to distance/similarity Examples of affinity measures

Euclidean

Manhattan

Hamming

L

iii AgAbD

1

2)(

L

iii AgAbD

1

L

i

ii AgAbD

1 otherwise0

if1δwhereδ,


Immunocomputing

Algorithms and Processes Generic algorithms based on specific

immune principles, processes or theoretical models

Main Types Bone marrow algorithms Thymus algorithms Clonal selection algorithms Immune network models


Exemple of application:

Immunocomputing


Another example of application:

Immunocomputing


Why immunocomputing? Adaptability Robustness Distributivity Decentralization Fault detection and tolerance Self/Nonself discrimination* ...

Immunocomputing


Scope: Pattern recognition Fault and anomaly detection, and the security

of information systems Data analysis (knowledge discovery in

databases, clustering, etc.) Agent-based systems Scheduling Machine-learning Autonomous navigation and control Search and optimization problems Artificial life

Immunocomputing

Part III-B

Artificial Life and Fractal GeometryArtificial Life and Fractal Geometry


Main Ideas

Biosciences: reductionist approach to understanding life

Artificial Life & Fractal Geometry: bottom-up approach to synthesize life patterns and behaviors

Focus on the computational synthesis of natural patterns and behaviors, not problem solving

Widely used in computer graphics and movie making


Artificial Life

What is life? “The property or quality that distinguishes

living organisms from dead organisms and inanimate matter, manifested in functions such as metabolism, growth, reproduction, and response to stimuli or adaptation to the environment originating from within the organism.” (Dictionary.com)

Are mules alive?


Some poetical definitions of life “Life is a long process of getting tired”

(Samuel Butler) “Life is a tale told by an idiot - full of sound

and fury, signifying nothing” (Shakespeare)

Artificial Life


Artificial Life: “Artificial Life is the study of man-made systems

that exhibit behaviors characteristic of natural living systems. It complements the traditional biological sciences concerned with the analysis of living organisms by attempting to synthesize life-like behaviors within computers and other artificial media. By extending the empirical foundation upon which biology is based beyond the carbon-chain life that has evolved on Earth, Artificial Life can contribute to theoretical biology by locating life-as-we-know-it within the larger picture of life-as-it-could-be.” (Chris Langton)

Artificial Life


“Artificial Life (AL) is the enterprise of understanding biology by constructing biological phenomena out of artificial components, rather than breaking natural life forms down into their component parts. It is the synthetic rather than the reductionist approach.” (Ray, 1994)

Artificial Life


“Alife is a constructive endeavor: Some researchers aim at evolving patterns in a computer; some seek to elicit social behaviors in real-world robots; others wish to study life-related phenomena in a more controllable setting, while still others are interested in the synthesis of novel lifelike systems in chemical, electronic, mechanical, and other artificial media. Alife is an experimental discipline, fundamentally consisting of the observation of run-time behaviors, those complex interactions generated when populations of man-made, artificial creatures are immersed in real or simulated environments.” (Ronald et al., 1999)

Artificial Life


Artificial Life

Natural Life: An instance


Boids: Simple Behavioral Rules Collision avoidance and separation Velocity match and alignment Flock centering or cohesion

Artificial Life


Boids

Artificial Life


AIBO ERS 210

Artificial Life


Artificial Life


Wasp Nest Building

Artificial Life


Creatures: Adaptive learning through interaction

Artificial Life


Artificial fishes: Predator behavior

Artificial Life


Traffic simulation: What is needed for a jam?

Artificial Life


Life-as-it-is x life-as-it-could-be

Artificial Life


Why Artificial Life? Increases our understanding of life Provides new perspectives about ‘life’ and its

many models Development of new technologies: softwares,

robotics, interactive games, computer graphics, educational systems, behavior animation tools

...

Artificial Life


Fractal Geometry

“Why is geometry often described as ‘cold’ and ‘dry’? One reason lies in its inability to describe the shape of a cloud, a mountain, a coastline, or a tree. Clouds are not spheres, mountains are not cones, coastlines are not circles, and bark is not smooth, nor does lightning travel in a straight line. … The existence of these patterns challenges us to study those forms that Euclid leaves aside as being ‘formless’, to investigate the morphology of the ‘amorphous’.” (Mandelbrot, 1983; p. 1)

A major breakthrough in the process of modeling and synthesizing natural patterns and structures was the recognition that nature is fractal and the development of fractal geometry

Fractal geometry is the geometry of nature with all its irregular, fragmented and complex structures


Fractal Geometry

Some Tools: Cellular automata Iterated function systems Lindenmayer systems Brownian motion Particle systems Evolutionary design etc.


Cellular automata Dynamical system that is discrete in both space

and time Prototypical models for complex systems and

processes consisting of a large number of identical, simple, locally interacting components

Formal description: C = (S,s0,G,d,f), S is a finite set of states, s0 S are the initial states of the CA, G is the cellular neighborhood, d Z+ is the dimension of C, and f is the local cellular interaction rule, also referred to

as the transition function or transition rule.

Fractal Geometry


Cellular automata

Fractal Geometry


Lindenmayer Systems A formalism to simulate the development of

multicellular organisms A string or word OL-system is defined as the

ordered triplet G = V,,P, where V is the alphabet of the system, V+ is a nonempty word called the axiom, and P V V* is a finite set of productions

The geometric interpretation of the words generated by an L-system can be used to generate schematic images of diverse natural patterns

Fractal Geometry


Lindenmayer Systems (without rendering)

Fractal Geometry


Lindenmayer Systems (with rendering)

Fractal Geometry


Fractal Geometry

A Natural Fern


(Random) Iterated Function Systems An iterated function system (IFS) consists of a

complete metric space (X,d) together with a finite set of contraction mappings wn : X X, with respective contractivity factors sn, n = 1,2,…N.

Let {X; w1, w2,…, wN} be an IFS, where a probability pi > 0 has been assigned to each wi, i = 1,…,N, i pi = 1

Choose a point x X and then choose recursively and independently a new point x obtained by applying only one of the transformations, chosen according to a given probability, to the current point x

Fractal Geometry


A Fern Generated with a RIFS

Fractal Geometry


Brownian Motion To model some natural sceneries, it is

necessary to have curves that look different when magnified but still possess the same characteristic impression

The term fractional Brownian motion (fBm) was introduced to refer to a family of Gaussian random functions capable of providing useful models of various natural time series

Fractal Geometry


Fractional Brownian Motion (without rendering)

Fractal Geometry


Fractional Brownian Motion (with rendering)

Fractal Geometry


Particle Systems Modeling physical phenomena like the flowing,

dripping and pouring of liquids, the liquid mixing with other substances, gases in motion, explosions, clouds, fireworks, etc.

A particle system consists of a collection of particles (objects) with various properties and some behavioral rules they must obey

The precise definition of these properties and laws depends on what is intended to be modeled

Fractal Geometry


Fractal Geometry

Particle Systems See

http://www.cs.wpi.edu/~matt/courses/cs563/talks/psys.html


Why fractal geometry? A computationally cheap way of generating

computer models of nature Study natural patterns: extinct vegetation,

design new variety of plants, study growth and developmental processes, aid farmers and decorators, crop prediction, computer graphics and movie making, etc.

Fractal Geometry

Part III-C

Computing with New Natural Computing with New Natural MaterialMaterial


If current computing technology will reach its limit in the near future, what would be the alternative material with which to compute?

New computing methods based on other natural material than silicon: Molecules Membranes Quantum elements

Computing with Natural Material



DNA Computing


Quantum Computing Quantum bit: |x = c1|0 + c2|1


Part IV

Computing in the New MillenniumComputing in the New Millennium


Some ideas that form the basis of natural computing:

Capacity of dealing with complex problems The use of sets of candidate solutions Capacity of dealing imprecisely with imprecise

information Robustness Distributivity Self-repair etc.

Computing in the New Millennium



From singularity to plurality


The importance of nature has never been so great!



Main Reference

Fundamentals of Natural Computing, Concepts, Algorithms, and Applications; by Leandro de Castro, CRC Press, 2006


How far can we go?

Questions, comments?

2005: natural computing - concepts and applications

Technology

fractional

solve complex

natural phenomena

part iii

collective

systems based

computer graphics

natural computing