Intro to Boosting
Boosting is a simple concept which coincidentally prides itself on simplicity. The idea can be found in looking at the very root principle of artificial intelligence itself – that is, a machine learning to make accurate predictions based on past observations. Boosting follows this exact concept. By implementing several “weak” or “base” learners (which are basically just very simplistic observations made by the machine), the machine is able to better carry out its function. When fully put to good use, boosting using several of these weak learners can actually be in effect more efficient than implementing a single “strong” learner – a more complex set of instructions that helps the computer “learn” how to better itself in a similar way but is much more thorough than any weak learner could ever be.
However boosting is probably best described through use of an example, and the most recent, relatable, convenient and original one I could use would be facebook’s chatting software. The last large-scale update facebook made (or at least the recent one I cared about most) was a revamping of their newsfeed display as well as their chatting software. I didn’t really appreciate the fact that friends no longer had their picture display (with which I would usually identify someone immediately), but what struck me as interesting was the fact that friends most frequently conversed with(and their pictures) were displayed on the left side of the screen in a sort of box-square-like structure. This to me is a perfect example of boosting. Based on the number of lines, words, or times (or any combination of these three things) you have conversed with someone relative to the other people that are online, certain people have priority in the order in which they are displayed in this box. The number of lines, words, or instances of conversation would all be weak learners that stack up over time to create a value that would help facebook to determine who currently on the site is actually your most-preferred person to converse with. This in turn makes it much easier to find friends you actually want to talk to, as opposed to the other 500 people you added just for the sake of boredom.
(I’m not exactly sure if facebook actually implemented boosting, but in theory it is a good example.)
Sorry about the wiki post. I was told it would be fine to take things off wiki and I was not at the lab to actually ask about it.