Technically I'm a machine too. An electric meat machine.

September 8, 2011 8:35 AM

What's the difference between Stanford's CS221 and CS229 classes?

Stanford has two free classes this fall that look interesting: one in Machine Learning (CS229) and one in Artificial Intelligence (CS221).

I'd like to take one, but I'm not sure how to choose. There looks to be some overlap in material, but most of the terminology is foreign to me. What criteria can I use to pick? CS221 requires me to buy a textbook. CS229 does not.

posted by mkb to Education (6 answers total)

Have you considered talking to the professors of the courses and asking this question? They can almost certainly give you more information about what the courses are about and the differences between them.
posted by brainmouse at 8:44 AM on September 8, 2011

Machine learning is used in artificial intelligence but mostly covers the data input/processing/sorting side of things. Using Google's driverless cars as an example, artificial intelligence would be about creating a program that drives a car, and machine learning would be more about how Google gathers all that data from their cars and writes programs to automatically sort the data into something the car AI can use to make decisions with.

To use another example, a robot that automatically replies to and deletes e-mail is a simple AI, and Google's spam/not-spam Bayesian sorting is an example of machine learning.
posted by michaelh at 8:59 AM on September 8, 2011

Without any knowledge of either course forger than the titles I would say; take the ai course, a larger over view of the discipline will probably help you more than just ml. You can always read up on ml later if it interests you.
posted by adventureloop at 9:51 AM on September 8, 2011

If you have no knowledge of AI, take AI first. Machine Learning is just one subset of AI, and therefore much more specific. A basic amount of Machine Learning is covered in the AI course, actually.

Also, the AI course doesn't require the book, just recommends the book. And, if it is successful, I believe they are planning on offering these courses again in the future, so you will have another shot at it.
posted by semp at 10:01 AM on September 8, 2011

ML is basically about inductive reasoning— trying to derive information, generalities or patterns from sets of data. Machine learning can be used in non-AI contexts, and AI does not necessarily involve any machine learning. But if the reason you're interested in the ML course is that you have a broader interest in AI, then take the AI course first.
posted by hattifattener at 11:16 AM on September 8, 2011

What are you hoping to learn? What about the courses sounds interesting to you? I assume that youre taking the online courses, not the in-person Stanford class, and thus have no access to the instructors.

Without knowing anything more about you, count me as another vote for taking AI first. ML is massively useful, but not nearly as broad a course as the AI one. As an undergrad, I was a bit disappointed by my first ML course ... The name seems so exciting, but it winds up being how to classify data. An intro AI course will deal with problems that have a more CS/algorithmic flavor, while ML feels more like statistics to me.

(I have only glanced at the course outlines, and have no familiarity with those specific courses. I have taken the equivalent grad-level courses at a different university)
posted by Metasyntactic at 3:19 PM on September 8, 2011