Sections  Introduction  Emergence  Complexity  Chaos Thoery  Cellular Automata  Life Forms  Neural Networks   Evolution  L systesm  Uses people.htmlBooks Links about this site Site Index  Front Page
Stewart Dean's Guide to Artificial Life

The Uses and Abuses of ALife

(N.B. - this page needs input - please tell me about any applications you think should be mentioned.)

Alife is, as John Conway would describe, a big box of flashing lights that sooner or later everyone will find a use for. Many uses have already been found. In terms of computer aided design genetic algorithms have already played a large part in air craft design.

Personal Assistants

.Information agents.

Some people now say we live in an 'information society', or a 'telematic society' of even an 'Infoglut age'. Many different people have painted different pictures of what is agreed upon as being an information full future, or at least nearly agreed upon. Theodore Roszak in The Cult of Information (1986) claims that the move toward and information age is in a 'conspiracy by the Pentagon, vested interests, multinational and academics. He correctly pinpointed some of the main influences but to say it was a conspiracy may be harsh. People like having information at their finger tips, it is a form of empowerment.

Ninety eighty-four has been and gone and it is true to say that what we now face is far from the restrictions of thought enforced by newspeak but the problems of to much information. Information is constantly thrown at us from television, magazines, billboards and now the Internet. In Robert Heller's "Culture Shock", a book aimed at office management, there is a section on information overload. The introduction says.

Information systems only fully justify their cost when they provide strategic value. There is a crucial difference between benefiting fully from intelligence - by converting it into understanding - and merely receiving information.

This problem is only going to get worse if we continue to handle data as if it was information and do not start using artificial forms of filter to increase the richness of the information we receive. Alife has a part to play here as filters that simply stop data passing through will not be effective. What is needed is intelligent data processing, the results could prevent many headaches and save infonauts all over the world from dying from stress.

A guide to Information theory (Data is not Information)

What may be helpful to point out is the difference between data and information, a difference at the heart of information theory, but is often confused in everyday use. Data is what information is made out of. It is numbers, words, phrases and so on. These items can only be termed information when coupled with context and made relevant. Information in its pure form is rarely the problem in information overload, instead free floating data or unrelated information is the problem.

One of the best examples of information handling exists in the most complex computer we currently know, our own minds (that is us human blowing our own trumpet, see the section on neural networks to see that this will not always be the case). Our vision system is level upon level of filters that attempts to get rid of data and just provide information. That is we do not 'see' every thing our eyes receive. For example the receptors of the eyes mostly pick up change. This may seam strange until you realise that once a piece of visual data is received and processed there is no real need to send the same piece of data again, merely tell the brain when something changes and the resulting information will be the same.

This simple filter cuts down the data and increases the information. A similar technique is used in standard compression techniques on computers. The data that makes up a program or file is run length encoded, that is a string of the same number is replaced by the number and information about how many times that number happens. The data is 16 16 16 16 16 16 16, the information held by that data is seven sixteens.

This technique can also be applied to patterns. Patterns are currently hard for computers to sort out due to the complexity involved. Patterns are rarely obvious and writing a computer program to handle more than simple repetitive patterns is far from easy. We are very good at sorting out patterns, for example music. Music consists of many patterns, there are harmonies, rhythms, bars, choruses, movements, tunes as well as different timbres and even words as well. We can pick out most of the information out of what boils down to be a simple wave of amplitude over time. Therefore existing computer compression techniques are still run length encoding. Other forms of compression which are usable for graphics and sound, aesthetic data exist. These are lossy, that is some of the information is lost in the compression. The most well know are JPEG and MPEG that can produce high information compactness without loosing to much quality. It is possible to get well under one tenth of the size of the original with comparative quality. Lower rates start occurring what is known as low signal to noise ratios.

The less information extractable from a piece of data the higher the noise ratio. With certain applications some information can easily be thrown away, as with JPEG and MPEG. For example television pictures are made out of a high quality black and white picture with colour smeared on top. Colour is less important for a clear picture than a detailed monochrome image. Telephone lines use information theory to fit voice information down the line. This was a problem with early phone lines so it was worked out how much bandwidth was required for an understandable phone call and the rest was thrown away. The same is true of new the digital music formats, mini disc and digital compact cassette (DCC). The information the ear is said to not need is thrown away to fit more music in a smaller place.

The secret to good compression is therefore being able to work out how to get the information out of the data recreate the data at the other end. Currently the most advanced way of doing this is to use fractal compression. Using heavy computation patterns are squeezed out of pictures and stored as templates which are then used to reconstitute the picture at the other end. This technique will even fill in gaps in a picture if a section is missing, although this may not work very well on things such as faces.

Bringing information to life (or Alife to information).

To explain why this is relevant this is an example of the extraction of information from data to produce concentrated information. The original data could be said to have contained much redundant information and, more importantly, noise and so it is with the data we receive each day. The more information rich the data we receive the less data we have to sort through our selves. Sorting out data is a computationally high task. This applies to computers as much as people. The problem is enabling computers to be able to do this and, unsurprisingly, this is where aLife comes in.

What it is possible to do is create what are known as information agents which roam around 'cyberspace' making connections and working out what is relevant to them and what is not. Information agents are more commonly known as autonomous agents but here information agent is a better description. An agent is a single minded piece of code that sifts through data looking for references important to it and storing links with weights. In a system the piece of information should also keep a record of which agents it has links with, this will be explained. The more important or relevant this piece of semi intelligent code considers a file, the stronger it will make a link.

On its own a single agent is not much good as it is to narrow minded. Make it a broad-minded creature and it will latch on to anything and you end up with the problem of a lot of data and not much information. The solution is to send out a whole colony of agents that travel around competing and co-operating, much like bees and ants in a colony. Each one makes links with files and competes to have their weighting top of that files list. This is balanced by agents making links to other agents which share some common links. The result is the information becomes interconnected automatically, the more you run the agents, the more ordered the will get.

The environment these agents inhabit may be a single hard disk or the length and breadth of the Internet. As long as both sides keep the deal going (the file or cell of information and the kinetic agent) information becomes self organising. The result being automatic links. For a simplified version of this see the Microcosm project run by Portsmouth University. This PC based system also has basic techniques for dealing with other forms of multimedia information, mostly based around human labeling. Artificial life's forms of intelligence can also be applied to the task of evaluating data via speech recognition, perceptrons looking at pictures and intelligent parsers 'reading' text to find relevancy where simple word searches would fail. Each agent may have access to a whole tool kit of routines which it uses to analyse any clump of data it comes across.

The agents need not take up a lot of space, all that is needed is the information of what where and who, all the routines would be shared by each agent so only need exist once. Each agent would be a virtual creature. It may even be possible to bread creatures. If one creature is overwhelmed by links then it may spawn a copy (or breed? See evolution). It may even split in two, each with a different half of the links. Agents with few links may die to preserve computing time.

Variations from creature to creature may be important, as in nature. For the searching technique used by a creature to browse for information may be different for different classes of agent. Possible browsing techniques include:

  • Scanning - Covering a large area without depth; dipping into files.
  • Browsing - Following a path until a goal is reached, which may be less exact than searching. The link may not be the strongest but just the first above a certain value come across.
  • Searching - Striving to find a piece of information, following each strongest link till a dead end is reached.
  • Exploring - Searching several branches to find the extent of a piece of information.
  • Wandering. - Aimless link following picking up relevant pieces on the way.
A combination of these techniques would be used for a colony of agents to be doing the best job, as with an ant colony.

All that you would need to do is say which information you needed and the links would already be in place, meaning you would not have to search through reams of documents, reels of movies and double albums of sound to find what you wanted. The added advantage is any new document would soon be latched upon by these agents and place it in the correct portion of the web. This technique is not so much a filter as a means of indexing pre-processed data. It is guaranteed to save a lot of time and effectively give life to information.

Entertainment

Neural nets have already found their way into computer games in the quest for more convincing interaction. Whether the character or creature you are interacting with is a friend or foe a game or piece of interaction is far more interesting if that character behaves in an intelligent way. Uses for this include realistic adventure games, smart opponents for sports or board games and tricky aliens to blast. Direct uses have been found in games like SimLife (yet another simulation game) and more recently 'Creatures'.

It would also be possible to create presenter that adapted to varying circumstances and interact with the audience realistically. The concept of presenters of guides has existed for some time but has yet to be used with great impact. As electronic entertainment media becomes more advanced so demands for more realism will grow. Without the techniques of artificial art this may be difficult. allows William to produce sculptures simply not possible in real life. The organic like forms created are a mixture of varying techniques, many of which have much to do with aLife. Using evolution organic structures can be grown on the screen.

Also examples of human selected genetic art can be found on the net. Using the principle of genetic algorithms the pictures evolve over time and are scored according to their aesthetic worth. The idea is art works are produce democratically, emergent are if you like.

Alife can also be used in image processing and offers interesting possibilities of cybernetic artists. Rave culture has already embraced the ideas of aLife, if only the sound of them and terms associated with artificial life often appear in the club culture.

Music to is increasingly being 'grown' in much the same way as art. Unfortunately often the results of a bad piece of music can be a lot worse than a bad picture. Most people like harmony and tunes and genetic music can at time be a cacophony and resemble avant-garde music which is hard to digest and is often thought of as far from beautiful. Arguably music full of maths and there is no reason why computers should not get better at making music. Whether this is acceptable by the purists is arguable. There are far more music purists than artistic purists, but the possibilities exist.

War

It is an unfortunate thing that some people still think that barriers, and more disturbing, wars are a good idea. Although the Internet may stratify society along new lines in the future it is also hoped that the world will seem a much smaller place and barriers suddenly become less important. Whatever the reality or the hoped for we still live in a world where people will die for a cause. As there are many different causes enshrouded with history, propaganda and culture war is inevitable.

I have already indicated how powerful aLife is and can be so it is unsurprising that the military has a strong interest in aLife. In Maneul De Landa's book, War in the Age of Intelligent Machines, the history of war is taken into the future with the creation of autonomous weapons. It appears that the Defense Advanced Research Projects Agency of America has already got devices on its drawing board that takes the people out of combat or turns a single soldier into a battalion. Talk of cybernetic creature, merging flesh with metal and plastic are far from uncommon.

The military have always had a history of research that is kept put of the lime light and advanced. Alife should be about creation, about life. Within all possibilities artificial life research may have already been used to kill. This is not the result of technology, just the result of giving an egotistical ape a bigger stone to bash the apes from another tribe.

Viruses (It's aLife Jim, but not as we know it.)

Much talk has gone on about whether computer viruses are alive in the true sense of the world. One thing is certain, most of them mess up your computer in some way or other. The history of viruses is quite interesting. Most of it is centered around eastern Europe where many of the best known viruses were written. Some viruses were written for semi-political reasons, anarchists for example but most are written for sheer ego reasons. Viruses most often are programmed to flash messages from the creator or include as their code messages.

Some viruses have been found to have slightly mutated which have changed the way they worked, this is probably due to an error in a disk being read. An example of this is the Cascade virus, and old virus which causes letters to fall down the screen. The original would only infect a file once, like many other viruses. This stops the files growing beyond a usable size so therefore stopping the virus spreading eventually. The mutated version has one altered byte which allows it to write to files repeatedly.

As one of a computer viruses' aims in life is to reproduce and given that mutation can happen it is possible that a virus may evolve. The only hindering factors are new virus beating programs such as 'F-prot' on the PC and the slowness in mutation. If a virus was designed to mutate or even breed then this might change the situation. This may not stop antiviral programs wiping them out but may lead to strains that at least try and stay one step ahead of virus scanning technology.

Stephen Hawkings is in no doubt that computer viruses are alive and has even said that it is a shame that the first form of life we have created has been a malicious one, we have created a new life form in our own image.



Information Society

This term was reportedly first used in Kenichi Kohyama's 'Introduction to Information Society Theory' in 1968.



JPEG and MPEG

JPEG stands for Joint Photographic Experts Group and MPEG stands of Moving Picture Experts Group. These are the names of the oragnisations that devised these standard. Both really upon lossy techniques, that is compression that throws away information, to reproduce pictures and sound. With MPEG more can be thrown away in each frame due to the extra information often implied by movement.



Please send any corrections, comments or additions to: alife(at)stewdean(dot)com
 

 back a bit  next bit