So, we are still getting stuck into the maths this week, moving onto multivariate linear regression and some linear algebra stuff. It still just looks like stuff I did more than 20 years ago to solve tricky maths problems – done by dumb machines that happen to be able to add numbers up very quickly. It was good to crank up MatLab and try to figure out applying the fancy formulas with Greek letters into code that a machine could use – and also work stuff out very quickly.
But so far, as expected, no sign of SkyNet!
I am looking forward to the parts of the course – although maybe they will only be on more advanced courses – when aspects of logic, philosophy and learning start coming into play. It strikes me that the breakthrough in machine learning (and AI in general) when we can program machines to learn by themselves with no great structure or guidance being given – just the desire to learn, some examples and a guiding principles – kind of like what any kid with at least one concerned adult comes into the world with.
So far, it is all very mechanical. At least for now, that sort of thing seems very far away. Perhaps even in a galaxy far, far away…
But I’ll keep you posted.