Entries from September 2007 ↓

I’d like to present a motivation behind why some people might take drugs. Often, when prompted for a reason why, people might say something like “To expand my consciousness” or “To discover something about myself”. Of course, this does somewhat depend on the drug and who you ask, a lot of people just like them for the thrill or the immediate sensations. I’d like to explore the former reasons however.

Optimisation techniques

Since I’m a programmer, and enthusiast about artificial intelligence, I’m going to approach it from this angle. Particularly genetic-algorithms, simulated-annealing, neural-networks, and other optimisation techniques that have a solution space that one can visualise as being a rugged fitness landscape of peaks and troughs.

The height of a point on this landscape indicates the fitness of being at that particular point. Imagine you are standing at said point. If you move slightly in one direction it may increase your fitness, decrease it, or it may stay the same. All the above machine learning methods, in some way, are moving along a multi-dimensional landscape of fitness, all are trying to reach the peak fitness value. The problem however, is that, generally these methods only move in the direction of increasing fitness (although the specifics may be different). If you find yourself at the top of a peak, you’ve reached the locally optimal solution, but you’ve no way of knowing if you’ve reached the globally optimal solution.

Continue reading →

I'd like to present a motivation behind why some people might take drugs. Often, when prompted for a reason why, people might say something like "To expand my consciousness" or "To discover something about myself". Of course, this does somewhat depend on the drug and who you ask, a lot of people just like them for the thrill or the immediate sensations. I'd like to explore the former reasons however. Optimisation techniques Since I'm a programmer, and enthusiast about artificial intelligence, I'm going to approach it from this angle. Particularly genetic-algorithms, simulated-annealing, neural-networks, and other optimisation techniques that have a solution space that one can visualise as being a rugged fitness landscape of peaks and troughs. The height of a point on this landscape indicates the fitness of being at that particular point. Imagine you are standing at said point. If you move slightly in one direction it may increase your fitness, decrease it, or it may stay the same. All the above machine learning methods, in some way, are moving along a multi-dimensional landscape of fitness, all are trying to reach the peak fitness value. The problem however, is that, generally these methods only move in the direction of increasing fitness (although the specifics may be different). If you find yourself at the top of a peak, you've reached the locally optimal solution, but you've no way of knowing if you've reached the globally optimal solution. The way these optimisation techniques get good results however, is by including an element of stochasticity or randomness. Either this randomness is added to the path that the solution point takes along the landscape, such that it will sometimes move down slopes, or suddenly jump onto another nearby hill. Or the optimisation is run many many times from different start points, and by chance, one of these start points is likely to be at the base of a peak that represents a pretty good solution, even if the solution isn't the best. Even including a small amount of this random behaviour can have a significant effect on the performance of these techniques. Obviously it depends on the problem you are trying to solve, and particularly how rugged it's fitness landscape is. If the landscape is just one huge peak, then it doesn't matter where you start, you'll get to the best solution eventually. Storing patterns in the mind This is where I start getting beyond my formal training, and I'm relying on various books I've read, but can't remember where or when, so this could be me just spouting mumbo jumbo. You've been warned. I read somewhere that one of the leading ideas about the purpose of dreams, and sleep in general, is to find a low energy state in which to store information. Thus, when you're asleep, your brain is trying to juggle all the data that is running around in your active, short term memory and meld it with the existing neural pathways. This is another optimisation problem, based on all your existing memories and knowledge, which are stored in intricate patterns and interleave one another, how do you include the new memories while using the least energy. Or put another way, what is the most efficient way to store the new experiences and information you've received since you last slept? This information is probably weighted by how important it is in your day and such importance may be described by either the emotional response it invokes or through sheer repetition (which is how neural pathways are strengthened). The latter explains why if you do something monotonous all day, even if it's of little consequence to your life, you'll often end up dreaming about it anyway. The long term result Since memories are being layered over our existing neural pathways, it isn't causing significant change. Unless of course something major happens in life which requires adaptation. We can easily get stuck in our ways, or in our thought routines, since the patterns that build up while storing memories are the same patterns that result in our intelligence, behaviour, and conscious existence. Often people think about intelligence and memory as separate entities, and I very much doubt they are. Memory are patterns in your mind, just like your intelligent thought. I admit however that certain areas of the brain are dedicated to the storage of particular types of knowledge, but the use of this knowledge is thought, and without the thoughts activating the knowledge it may as well not exist. Anyhow, basically what I'm trying to say is that our mind, as it consolidates knowledge, memories, and our experience, can get stuck on a hill on the fitness landscape... not only for the lowest energy state, but also for the best interpretation on the knowledge that has been integrated into our minds, and the most satisfying direction in life. It's easy to get stuck in a rut. Drugs shake it up Why I think people take certain drugs, is that it distorts the fitness landscape for whatever optimisation problem is being run by the human mind (I don't really think intelligence is as simple as an optimisation process, it's just for the purposes of this argument). It's like the stochasticity that machine learning methods use to explore hills in the fitness landscape that are nearby and possibly have better overall fitness. Many recreational drugs have had a past as an aid in psychotherapy. I think this might be why. But then again, a drug's effects could lead to a less fit peak for the brain to climb, thus for some people drugs have a negative impact. How do you know what result to expect? I guess you don't, but it's probably safe to say that some drugs are more likely to result in the net benefit to you, while others are not. I recommend Erowid if you want information about a recreational/nootropic drug. Disclaimer: I am not condoning drug use, or otherwise. These are just my musings.

So BarCampChristchurch went pretty well. Quite a few interesting talks/discussions and I think my talk on the Singularity held some interest for people:

It’s also fortuitous that the BarCamp was on the Friday before the Singularity Summit 2007 because it meant I could at least pretend I was attending! On a side note, the Summit got on the front page of the San Francisco Chronicle.

Ben Kepes and Marek Kuziel live-blogged the event. There were even some rumours of video being recorded. I’ll make an update if that eventuates (and I don’t look like a daft fool).

I enjoyed going out for beer and Indian dinner afterwards. Met a bunch of interesting web types, which was cool because I’ve been somewhat isolated from the web-tech world while doing my PhD in an Ecology department. I also heard some more about The Valley In Christchurch meals, which I’d like to attend… once I’ve exumed myself from the monetary grave I am in right now!

So BarCampChristchurch went pretty well. Quite a few interesting talks/discussions and I think my talk on the Singularity held some interest for people: It's also fortuitous that the BarCamp was on the Friday before the Singularity Summit 2007 because it meant I could at least pretend I was attending! On a side note, the Summit got on the front page of the San Francisco Chronicle. Ben Kepes and Marek Kuziel live-blogged the event. There were even some rumours of video being recorded. I'll make an update if that eventuates (and I don't look like a daft fool). I enjoyed going out for beer and Indian dinner afterwards. Met a bunch of interesting web types, which was cool because I've been somewhat isolated from the web-tech world while doing my PhD in an Ecology department. I also heard some more about The Valley In Christchurch meals, which I'd like to attend... once I've exumed myself from the monetary grave I am in right now!

A whole group of friends from Wellington and a couple from Australia came to Christchurch recently. For snowboarding, and for our alternate event to Burning Man…. Melting Man. Helix DJed at Nitrate, with some of the WDC there to support him on the Saturday. We travelled to Flock Hill station on Sunday, and on Monday I mountain biked like a crazy man (on the Broken Hill track). Monday evening was building a snowman, dancing on a hill, and having good times (more about it from Kathy and Tatjna).

A whole group of friends from Wellington and a couple from Australia came to Christchurch recently. For snowboarding, and for our alternate event to Burning Man.... Melting Man. Helix DJed at Nitrate, with some of the WDC there to support him on the Saturday. We travelled to Flock Hill station on Sunday, and on Monday I mountain biked like a crazy man (on the Broken Hill track). Monday evening was building a snowman, dancing on a hill, and having good times (more about it from Kathy and Tatjna).