Keeping the zoo...
Jan. 9th, 2004 02:10 amSo for cs4611 with Ashok, we're playing with a particular sort of system that knows some general rules and has a short-term working memory... which could be a lot simpler than you'd think, but it's really powerful, and it at least feels like how a human would think about things...
What makes this different from just a big string of "if-then" code is that you've got some clever datastructure to hold all of your rules, and you need some clever way to search through it to find what you're trying to apply, because not all of them are going to apply.
So in the reading, there was this entertaining toy system with a fun background story: Robbie the robot wants to go to the zoo for the afternoon, and while he's there, he decides to try to classify all the animals he sees... so he codes up this system for himself that does inference (deduction, specifically) to figure out what animals are what, based on his percepts.
So I wrote that... just a good exercise. This made me feel a little better about Ashok's class, which has been getting slightly threateningly theoretical -- the sort of thing where I was thinking "eh, that intuitively makes sense, but I wonder if I could ever code it..."
That's what I love about CS, really -- the background stories. AI is all about good background stories... (physics is fun when it does that, as well...)
What makes this different from just a big string of "if-then" code is that you've got some clever datastructure to hold all of your rules, and you need some clever way to search through it to find what you're trying to apply, because not all of them are going to apply.
So in the reading, there was this entertaining toy system with a fun background story: Robbie the robot wants to go to the zoo for the afternoon, and while he's there, he decides to try to classify all the animals he sees... so he codes up this system for himself that does inference (deduction, specifically) to figure out what animals are what, based on his percepts.
So I wrote that... just a good exercise. This made me feel a little better about Ashok's class, which has been getting slightly threateningly theoretical -- the sort of thing where I was thinking "eh, that intuitively makes sense, but I wonder if I could ever code it..."
That's what I love about CS, really -- the background stories. AI is all about good background stories... (physics is fun when it does that, as well...)