August 26, 2002
Someone set the sky on fire
This lovely picture is from a video fragment taken from my kitchen windows looking over to the sky above Grangemouth. Say what you like about pollution but we get some mighty fine sunsets where I live.
Updating as you may have noticed is still as eratic as ever - I'm getting on top of things and managing to get more done - its just a case of getting back into the swing of things on the website front. I have been mostly learning Scheme over the last couple of weeks. As part of a number of ongoing projects both my own (primarily the "abitlike..." project) and for work (~advantage) I am exploring probability, markov chains and bayesian networks and scheme is an ideal langauge to try out quite complex ideas very quickly as well as being fun to work with.
The reason behind my interest in probability is more complicated. I have always been interested in so called "Artificial Intelligence" - as a field it has a number of problems ( not least because some of its practitioners have been going round claiming we will build a human brain within the next 5 years since the '70s) but underneath the hype has provided a number of technologies that have proved very powerful indeed. Interestingly when technologies that work are produced by the AI field they rapidly stop being called AI. Anyway from studying this over the years and following research in the fields of psychology and cognitive science and other related areas I have come to the conclusion that the brain isn't the all singing all dancing super machine it is normally portrayed to be. As far as I can tell (from other peoples work...this isn't some incredible insight I have come up with myself you understand) the brain is a collection of heuristic and adaptive mechanisms with consciousness being a slow overview type process that steps in when the underlying mechanisms need help. Now that’s quite a simplistic view but it will do for now - the interesting point being the heuristic and adaptive sub processes. It looks like some very interesting results can be gotten using very probabilistic models in conjunction with heuristics which is exactly what I am playing with.
That wasn't very well written - oh well. There will be more entries on this as I get things working.
I've been reading Linked which while not quite as ground breaking as it makes out is and interesting and compelling read. The science of networks is one which seemed like a bit of a back water until the growth of the internet but turns out to be relevant almost everywhere there are relationships between things. It is well written (more interesting magazine article than textbook) and is full of interesting titbits - the origin of the six degrees of Kevin Bacon game and the maths behind it some of which was done by Stanley Milgram more famous for showing that people will torture other people if told to do so by other people in uniforms*. Highly recomended.
Music wise I've been having a bit of phase of interesting male singers - most particularly with the powerful tones of Mr John Cale, Peter Hammill's more difficult work and the utterly wonderful "tilt" by Scott Walker. I have also rediscovered Heldon". The new Peter Gabriel album is also very very good indeed and is a must buy when it comes out.
*note intentionaly over simplistic summary
Posted by Mark at August 26, 2002 12:12 PM