What my MSc is Actually About
I spend quite a lot of time explaining to people what the hell I’m actually studying. It’s quite difficult because my MSc is made up a motley collection of modules that don’t really have much in common.
The central theme is Natural Computation. It’s basically about
- Looking at the ways nature processes information- all kinds of information. Computers aren’t the only things that compute- animals, plants, cells and bacteria, anything that lives, has to make decisions and “think” in some way. Even if it’s as simplistic as angling a leaf to perfectly catch the light. (Which actually turns out to be very complicated indeed).
- Trying to learn from these processes and copy them in our electronic computers. Imitating nature often gives us far better results than trying to solve problems ourselves. Artificial computers are blazingly fast but incredibly stupid. Nature’s computers (the brain, the immune system, and so on) are much slower, but their complex architectures make them incredibly cunning.
My modules this term are…
- Complex Bio-Inspired Algorithms: we looked at how ants work together to find food, how the immune system decides what it should take out and what it should leave alone, and how birds and fish avoid predators by flocking. Then we saw how we can learn from these to build pathfinding algorithms (the ants), antivirus and fault detection systems (the immune system) and swarms of cool-looking but rather useless £400 doughnut-sized robots that can flock together- provided someone turns them around before they fall off the edge of the table.
- Neural Networks: we studied systems that copy the brain (all brains, human to insect, share roughly the same architecture). The human brain is an interlinked web of nearly a hundred billion cells called neurons, and the patterns of electrical current flowing between them are our thoughts. The largest artificial brain has 20 billion neurons and runs on a supercomputer, and it can’t do very much compared to a person. Yet.
- Evolutionary algorithms: your body is an evolved solution to the problem of keeping yourself alive. Over millions of years we’ve become more and more sophisticated and tuned as natural selection acted on our gene pool. Wouldn’t it be nice if we could solve real-world problems like traffic routing or even electronic circuit design just by letting the answers evolve for themselves? Well, we can. You have to generate a huge amount of random solutions, kill off the shit ones, keep the good ones, maybe mix things up a bit, and repeat. This process has been used to generate quite complex electronic circuits that do things like measure frequency. And we don’t understand how they work. And we don’t need to. They evolved themselves, and they work.
- Quantum information processing: I’m not sure why this module is even in my course, as it’s not very “mother nature”- it underlies the physics behind the entire universe, not just a bunch of furry things on a random planet. But I@m glad it is. It’s absolutely mind-blowing. Basically, on the subatomic level, things don’t really exist. Objects (such as they are) don’t have a definite location or state- just a set of states in which they might be found. Classical objects have being. Quantum objects exist only in superpositions of different states. By juggling these superpositions and bending them around (we usually work with superpositions of light and laser beams) we can build computers. They’re not just faster than classical computers. They’re on a whole new level. Quantum computing is going to be more revolutionary than the transistor or the processor. And quantum theory as a whole rips the foundations out from the entire way you think about the world.
It’s been rather fun so far.
Next term we’re doing…
- Emergent systems- large interacting systems with very simple rules that, when looked at as a whole, have very exciting properties.
- Simulating complex biosystems, like proteins, DNA and cells.
- Quantitative research methods- how to do experiments properly. Yes, PROPERLY. No more of this “oooh, method A is 0.0002% more effective than method B! REVOLUTION!!”
- Evolvable hardware. The self-building circuits that I was on about.
- Computing with biology and chemistry- using jars of DNA to work stuff out for you! Like the computers in Star Trek: Voyager!
Then I get a 6-month project as part of which I am apparently expected to publish papers. LOL.
