An intro to the world of AI
The minute you mention Artificial Intelligence people automatically think terminator! Well I’m here to tell you that given the current state of
affairs this is unlikely in the near future which may come as a relief to some and a disappointment to others. In order to fill you in on the current
state of affairs it is important to start at the beginning.
Our story begins in the 1970’s (roughly). This is when the idea of AI first came about and started to be worked on in academic fields. Much work was
carried out in the UK at places like Edinburgh, and one of the lectures there is considered to be the father of modern AI.
People can follow this link to view more information on the subject
www.inf.ed.ac.uk...
Back in the day all sorts of grandiose predictions were made as to how much progress would be made in the field. Indeed many people predicted that the
problem of intelligent computers would be solved in no time at all (look at Hal in 2001, even Mr C Clarke got it wrong! As an interesting side note,
my grandfather was once bitten by his dog, but I digress….). Oh how wrong they were. Having spent the last three years looking at their work, I can
say without a doubt that this is the most difficult thing we as a race will ever undertake.
Before the problem even begins to be looked at we need to define what intelligence is and that in itself takes us into a huge philosophical debate.
Rather than regurgitate what other people have said I suggest people look at the following links to get an idea of what computer scientists are up
against and what people have judged as goals for AI to achieve before scientists can pat themselves on the back and move onto other things.
www.akri.org...
www.cs.virginia.edu...
www.csee.usf.edu...
www.cse.buffalo.edu...
These days people have taken a divide and conquer approach to the world of AI. Rather than setting about making intelligent computers, the problem has
been subdivided into numerous categories and the result has been several spin off technologies which are now in use in different systems all over the
world.
The most interesting widespread in my opinion relate to the areas of:
• Neural Nets
• Genetic Algorithms
• Natural Language Processing (Syntax of language, how it all works)
• Computational Representations of Meaning (How we represent the semantics of language)
• Expert Systems (knowledge bases)
• Computer Vision (Image processing techniques and algorithms)
• Machine Learning (Techniques used to refine knowledge bases)
Neural Nets
Neural Nets are very useful little tools and while they start off simple they soon get a lot more complicated. In a nutshell the idea is to create a
system based fundamentally on a greatly simplified model of the human neurone. When linked together in nets these neurones allow us to perform
numerous tasks. The recommended reading for this module was a book by a man named Kevin Gurney. Several of my mates swore by this book and although at
times it is a little intense you will come away an expert by the time you finish it. Check out the following link and note that yes you can print out
his book for free, or alternatively read the slides online!
www.shef.ac.uk...
There are other people who talk about neural nets on the web but I have to say that this guy is the daddy. You need only bother reading this book if
you want to find out about neural nets.
Genetic Algorithms
These little blighters are algorithms based on the process of evolution and survival of the fittest. There are very, very simple to learn about and
rather interesting as a concept. A friend of mine once wrote a JAVA program to view them in action and found it to be a very interesting
experience.
cs.felk.cvut.cz...
www.rennard.org...
These links more or less tell you everything you need to know about when it comes to GA’s.
Natural Language Processing
I tell you people, if you write a program to crack this problem then you will become the richest person in the world overnight. This has so many
applications it is untrue. Sadly it is also very difficult and in my opinion modern techniques just aren’t up to it.
As far as topics go it is very vast. To get you started the best book to read is probably found on the following link:
www.cs.colorado.edu...
It written in a way that is easy to read for an academic book. It will also set you right on all of the current techniques for natural language
processing.
Computational Representations of Meaning
Its all very well been able to parse natural language and tag it with useful information which is what natural language processing deals with. I.e. it
deals with the syntax of language whereas this deals with the structure of language. This subject is so fringe and so little is understood about it
that it is difficult to find many useful sources on it. It is a very demanding subject area but very interesting from a philosophical point of view.
At UMIST this was taught by one of the most intelligent people I have ever met. He was a real genius and a leader in the field. (This man has more
degrees, doctorates, PhD’s than I have had hot meals!) A guy called Dr Alan Ramsey. At this time I’ve been unable to locate my notes for him so
I’ll follow with some sources later if I can find them. Besides as this is probably the hardest AI topic it makes sense that if you want to get
grounding in AI you read about the other topics first! Even so for those of you who are interested try looking at:
www.co.umist.ac.uk...
Expert Systems
These are used quite a lot in the medical profession and involve vast amounts of rules in the form of
The Left Hand Side -> Defined by the Right Hand Side
Or
A Cold -> Defined by a runny nose, headache, sneezes
These rules are then parsed and people can then query the rule base. As the above examples show these can be very useful and are created by a process
where an expert is sat down and rules are created via his knowledge. Much of my experience with these types of structures was done through the use of
a funny little language called PROLOG which stands for programmable logic.
www.amazon.co.uk...=1131723098/sr=1-1/ref=sr_1_3_1/202-0317851-4808660
This book will tell you everything you need to know. For those of you out there who are programmers, it is well worth giving Prolog a try. Its
completely different to C, JAVA or any other language out there.
Computer Vision
Another interesting topic, but if I were you guys I would concentrate on the others first as they are the most interesting. If your interested however
I would look at the following book as it is very good. (It got me through my exams!)
www.palgrave.com...
Machine Learning
If you value your sanity I wouldn’t bother with this subject. Its hella boring and I mean boring! You have been warned!
That’s all I can be bothered to write at the moment. Stay tuned folks!