9
$\begingroup$

I've always been interested in machine learning, but I can't figure out one thing about starting out with a simple "Hello World" example - how can I avoid hard-coding behavior?

For example, if I wanted to "teach" a bot how to avoid randomly placed obstacles, I couldn't just use relative motion, because the obstacles move around, but I don't want to hard code, say, distance, because that ruins the whole point of machine learning.

Obviously, randomly generating code would be impractical, so how could I do this?

$\endgroup$
1
  • 9
    $\begingroup$ this is a super theoretical AI question. An interesting discussion! but out of place... $\endgroup$ – Vass May 14 '14 at 0:23
5
$\begingroup$

Not sure if this fits the scope of this SE, but here's a stab at an answer anyway.

With all AI approaches you have to decide what it is you're modelling and what kind of uncertainty there is. Once you pick a framework that allows modelling of your situation, you then see which elements are "fixed" and which are flexible. For example, the model may allow you to define your own network structure (or even learn it) with certain constraints. You have to decide whether this flexibility is sufficient for your purposes. Then within a particular network structure, you can learn parameters given a specific training dataset.

You rarely hard-code behavior in AI/ML solutions. It's all about modelling the underlying situation and accommodating different situations by tweaking elements of the model.

In your example, perhaps you might have the robot learn how to detect obstacles (by analyzing elements in the environment), or you might have it keep track of where the obstacles were and which way they were moving.

$\endgroup$

Not the answer you're looking for? Browse other questions tagged or ask your own question.