Join 3,431 readers in helping fund MetaFilter (Hide)


Stanford online courses
November 22, 2011 7:54 AM   Subscribe

Stanford has announced new online courses for January 2012. Like the three courses currently running (1,2,3), these courses are free, open to the general public, and have no required textbook (previously).

CS 101
Design & Analysis of Algorithms I
Game Theory
Software Engineering for Software as a Service
Human-Computer Interaction
Cryptography
Probabilistic Graphical Models
Machine Learning (same course as currently running)
Natural Language Processing
The Lean Launchpad
Technology Entrepreneurship
posted by -jf- (28 comments total) 143 users marked this as a favorite

 
A recommended book for Natural Language Processing, by Manning, Raghavan, and Sch├╝tze, can be found here (this link is on the course page, too).

A recommended book for the algorithms class, by Dasgupta, Papadimitriou, and Vazirani, can be found here (pdf).
posted by -jf- at 7:55 AM on November 22, 2011 [4 favorites]


The "no required textbook" on the AI class was a bit of a joke. I made it through the first couple of weeks just fine and then suddenly found I didn't know enough prob&stats to follow along. The worked examples didn't provide enough information to figure it out, which means a textbook is only "not required" if you've already got everything they are going to use in your head.
posted by DU at 8:00 AM on November 22, 2011 [1 favorite]


Ooh, I hope that they offer these again next year when I'm out of school. I'd love to take...um, nearly all of these!
posted by smirkette at 8:08 AM on November 22, 2011


I think I might take the NLP one, especially if the textbook is online. The background requirements are a little intimidating but "length-normalizing a vector" is one of those things that sounds a lot harder than it is.

Not that CS isn't great, but would it kill them to do this for something like Vedic Texts or History of the USSR?
posted by theodolite at 8:16 AM on November 22, 2011 [1 favorite]


would it kill them to do this for something like Vedic Texts or History of the USSR?

Machine Learning + NLP + HCI ->* Teach your computer to teach you Vedic Texts and History of the USSR!

* Successful completion of this alchemy may lead to untold riches, Turing Awards, and/or being flamed on Slashdot.
posted by kmz at 8:25 AM on November 22, 2011 [3 favorites]


Is it possible to audit these classes? Do they post the videos publicly, or just to those signed up for the class?
posted by jsturgill at 8:27 AM on November 22, 2011


Is it possible to audit these classes? Do they post the videos publicly, or just to those signed up for the class?

If you sign up, you can do as little as you like. Nothing bad happens if you don't do the work.
posted by michaelh at 8:39 AM on November 22, 2011


I'm taking the Artificial Intelligence and Machine Learning classes now, and can recommend them both.

AI has ended up as a survey of current techniques. Since there aren't any programming assignments, any practical application will have to be done on your own. The homework assignments are good, but there's a big difference between knowing how to do something and actually writing the thing.

ML was a real sleeper. I was expecting it to be a dry, algorithm driven class but the combination of a great lecturer and engaging programming assignments has really kept my interest. I've been able to apply both classes in my day to day work, which has been satisfying.

Taking two classes has been much more work than I expected. I'm going to try to take two again next semester, but I would be struggling without the ability to spend some office time on the lectures (just ask!).
posted by (parenthetic me) at 8:43 AM on November 22, 2011 [1 favorite]


jsturgill: The classes are typically offered in a Basic and Advanced track. You do need to sign up for Basic to get access to the videos, but there's no downside except becoming part of their aggregate classroom analytics.

The AI videos are on YouTube, but it looks like next semester's batch of classes are using the ML class's platform. At least this time around, those aren't available without signing up.
posted by (parenthetic me) at 8:44 AM on November 22, 2011


I could definitely use a refresher in algorithms. Signing up now. I bounced out of the AI and ML classes before they even started since I just didn't have time to watch the videos!
posted by mkb at 8:50 AM on November 22, 2011


I lasted about a week doing the database one before I had to bail because while the class was very good (meatier and more demanding than a lot of online courses I've taken), it took too much time for me. I was also amused by the jabs at Berkeley in the videos.

I think these are some of the supporting materials online (they were probably there before the more recently publicized courses came about):

http://see.stanford.edu/see/courses.aspx

I remember seeing stuff for the database one too, but can't find that one atm.
posted by Seboshin at 8:50 AM on November 22, 2011


The last time I got to poking at free online courses, my family almost shot me over the "well, maybe one more career change..." they inspired.

But still, there's a lot of awesome stuff out there.
posted by gracedissolved at 8:59 AM on November 22, 2011


I lasted about a week doing the database one before I had to bail because while the class was very good (meatier and more demanding than a lot of online courses I've taken), it took too much time for me.

Same here. In truth, a full college class is just not something you can do in your spare time. You need to devote regular scheduled time, just like if it was a physical class where you had to go to a classroom twice a week.
posted by smackfu at 9:07 AM on November 22, 2011


Man, build one self-driving car and everyone wants to take your AI class.
posted by tommasz at 9:08 AM on November 22, 2011


You can get videos for some of these courses at Stanford's OpenClassroom site.
posted by -jf- at 9:35 AM on November 22, 2011


DU: "I made it through the first couple of weeks just fine and then suddenly found I didn't know enough prob&stats to follow along."

I find statistics to be so useful that I've picked up a textbook second hand for 7 bucks shipping included. I also keep Tanenbaum's Computer Networking textbook around, but you won't get that for a song.
posted by pwnguin at 9:38 AM on November 22, 2011


Hey, I was a TA for that NLP course once! If the online version is anything like the in-person version, it's a really interesting, challenging course.
posted by cider at 9:59 AM on November 22, 2011


I've also been taking the AI and ML classes and enjoying them a lot. Even though my math/stats background is really spotty, I haven't found them to be too hard. I do think the AI class is kind of inconsistent, maybe because it's being taught by two profs. There are moments of brilliance and other moments where the lectures seem to rush through a lot of material in a slapdash manner.

Sebastian Thrun's teaching style in particular is interesting -- on the (non-graded) quiz questions he'll actually ask for your intuition on material he hasn't taught yet. As a lover of puzzles I actually really enjoy this, but I can see how some people would find it alienating. It is very satisfying to successfully figure something out for yourself, though.

The ML class is practically flawless. The AI class is fun and challenging, but the ML class makes me feel like I really know the material, and as a very amateur programmer I've been surprised my ability to complete the programming assignments.

My only complaint about this new round of classes is that I have to restrain myself from taking them all, because they are a time commitment!
posted by speicus at 11:10 AM on November 22, 2011 [1 favorite]


it looks like next semester's batch of classes are using the ML class's platform

...which is great, because the ability to watch the lectures at 1.5x speed is AWESOME. I wish the classes I took in college had had that feature.
posted by Slothrup at 11:23 AM on November 22, 2011 [1 favorite]


If you have the comp. sic. and math background, you could obviously cover the same material in a fraction of the time by reading a book, but Andrew Ng has a wonderful lecturing style, even if he's slow enough that 1.5 speed becomes obligatory, and all the instructors have done an outstanding job imparting a sense of how the material actually applies in practice.

We should lobby for the condensed version, a paragraph summary of each video, with links to the videos.
posted by jeffburdges at 12:11 PM on November 22, 2011


Thanks!
posted by ZeusHumms at 1:42 PM on November 22, 2011


Andrew Ng has a wonderful lecturing style

Concretely!

I'm doing the DB and ML classes now, and they really are great. It's a serious time commitment, but I do feel like I'm getting a lot out of them. The worst part it trying to pick only two to do next semester. NLP, Game Theory, HCI, and Algorithms all look good.
posted by Who_Am_I at 2:56 PM on November 22, 2011


I am doing DB and AI. I am enjoying both, except that AI seems a bit of an experiment, while DB is much more polished.

The fact that the DB course's homework is structured around a series of workbenches where you can find out whether your solutions work is a big bonus. And Widom sounds like she is having a world of fun teaching this material, despite probably having taught it for years.

For both AI and DB I have found that having access to the textbooks helps, but it's not required. Google and the course's videos solve my questions 95% of the time.
posted by kandinski at 6:35 PM on November 22, 2011


> Sebastian Thrun's teaching style in particular is interesting -- on the (non-graded) quiz questions he'll actually ask for your intuition on material he hasn't taught yet. As a lover of puzzles I actually really enjoy this, but I can see how some people would find it alienating. It is very satisfying to successfully figure something out for yourself, though.

I love this style of teaching. The quizzes are mock-graded too (you get points that you can see, but which won't be in your final grade for the course), so your percentage of success tells you how good your previous knowledge or your intuition were. Have 100, you don't need to study so much. Have a 53, you know this is a unit you need to review.
posted by kandinski at 6:38 PM on November 22, 2011


Definitely take NLP! Far and away my favorite class at Stanford, and Chris Manning is an awesome teacher.
posted by troublesome at 10:57 PM on November 22, 2011


Apparently, they've added 3 more courses
Anatomy
Making Green Buildings
Information Theory
posted by -jf- at 2:30 PM on November 28, 2011 [1 favorite]


2 more
Model Thinking
Computer Security
posted by -jf- at 11:11 AM on November 29, 2011 [1 favorite]


MIT expands online learning and offers certificates
posted by jeffburdges at 12:52 PM on December 20, 2011


« Older Facebook has been criticized repeatedly for how it...  |  "The following HTML 5 movie co... Newer »


This thread has been archived and is closed to new comments