3 ms·

It's really a question of whether you prefer supervised learning or unsupervised learning. Supervised learning is one-off and expensive, unsupervised learning

by msg 18y ago

It's really a question of whether you prefer supervised learning or unsupervised learning. Supervised learning is one-off and expensive, unsupervised learning is cheap and widely available. Supervised learning converges over a much shorter time frame than unsupervised learning. Supervised learning introduces bias in the quality of the teacher signal that unsupervised learning cannot run afoul of (or take advantage of). An interesting middle way might be active learning. It is a technique where an unsupervised learner has the ability to ask a supervisor for useful tasks to train on. In academia, sadly, this doesn't seem to happen below the PhD level. http://en.wikipedia.org/wiki/Supervised_learning http://en.wikipedia.org/wiki/Supervised_learning http://en.wikipedia.org/wiki/Unsupervised_learning http://en.wikipedia.org/wiki/Unsupervised_learning

mechanical_fish 18y ago
* ...an unsupervised learner has the ability to ask a supervisor for useful tasks to train on. In academia, sadly, this doesn't seem to happen below the PhD level.* That isn't quite true. You're describing much of my high school math and physics education: I had teachers, and I "took" the official classes (i.e. I sat in class, most of the time, and I took the exams and did much of the homework...) but my teachers (who were quite awesome) recognized that I was ahead of the class and enjoyed teaching myself, so they threw me books like PSSC Physics and What is Calculus About? and told me to have fun. That was, however, unofficial and sort of under the radar. I expect such a style of learning to continue to rise in popularity now that the Web is here. It's just easier than ever before to run an active learning program: There's so much stuff online to steer the students toward, and even the offline stuff is easier to find -- there are book-recommendation services and blogs to help you find the best print books, and used-book services to sell them to you cheap. It really is getting to the point where I could set myself up as a "professor" right here in my home office and mentor folks over Skype as they go from one online resource to another, at big savings over paying a college for the same services.
- msg 18y ago
  Point taken. If I say your experience is the exception that proves the rule, though, I don't think that's too far off. I had a similar experience until midway through high school. I was about three years ahead of my peers in math, due to the efforts of one particular instructor who encouraged me to get ahead and stay ahead. After my ninth grade, he moved to another school. About the same time, my high school ran out of courses. I ended up taking throw-aways instead of vector calculus, differential equations, combinatorics, etc. for three years. I wonder if the push towards one-size-fits-all standardized testing has eliminated the possibility of making active learning official, at least in the public schools. Overworked teachers don't have the time to dumb it down and smart it up at the same time, whether the kids could handle it or not. There's a lot of generalizing here, sure, but I think the basic outline is true. Even my MS seminars were more teacher-directed than self-directed. Maybe I have a knack for choosing the wrong learning situations.

by msg 18y ago

mechanical_fish 18y ago

* ...an unsupervised learner has the ability to ask a supervisor for useful tasks to train on. In academia, sadly, this doesn't seem to happen below the PhD level.* That isn't quite true. You're describing much of my high school math and physics education: I had teachers, and I "took" the official classes (i.e. I sat in class, most of the time, and I took the exams and did much of the homework...) but my teachers (who were quite awesome) recognized that I was ahead of the class and enjoyed teaching myself, so they threw me books like PSSC Physics and What is Calculus About? and told me to have fun. That was, however, unofficial and sort of under the radar. I expect such a style of learning to continue to rise in popularity now that the Web is here. It's just easier than ever before to run an active learning program: There's so much stuff online to steer the students toward, and even the offline stuff is easier to find -- there are book-recommendation services and blogs to help you find the best print books, and used-book services to sell them to you cheap. It really is getting to the point where I could set myself up as a "professor" right here in my home office and mentor folks over Skype as they go from one online resource to another, at big savings over paying a college for the same services.

msg 18y ago
Point taken. If I say your experience is the exception that proves the rule, though, I don't think that's too far off. I had a similar experience until midway through high school. I was about three years ahead of my peers in math, due to the efforts of one particular instructor who encouraged me to get ahead and stay ahead. After my ninth grade, he moved to another school. About the same time, my high school ran out of courses. I ended up taking throw-aways instead of vector calculus, differential equations, combinatorics, etc. for three years. I wonder if the push towards one-size-fits-all standardized testing has eliminated the possibility of making active learning official, at least in the public schools. Overworked teachers don't have the time to dumb it down and smart it up at the same time, whether the kids could handle it or not. There's a lot of generalizing here, sure, but I think the basic outline is true. Even my MS seminars were more teacher-directed than self-directed. Maybe I have a knack for choosing the wrong learning situations.