Wow, does anyone know what is the system here? Has stanford just said to their faculty that this is something they can do in their spare time but stanford will not actively support it? Or is there some other arrangement going on?
Also why are they not shifting away from CS classes (except for entrepreneurship). Things like signal processing and basic controllers can be taught with such a class. I would specially love the signal processing class since there are 3 classes (don't exactly remember the name) which are considered holy perfecta of signal processing.
The service [1] described in that article allows authors to let people download an article from the ACM via. a link on their personal website, rather than having to host the PDF themselves. There's been no change in the access policy for non-members.
Any chance that they might do a compilers class too? The compilers class at my school got cancelled for the fall term and I don't think I'll have a chance to take it in a future term.
The introductory compilers class, CS143, will be available to Stanford students in the spring. I have no idea if they'll put it online, but it would be consistent with their current initiatives. I think you'll have to inquire more or wait until then to find out.
I work full time and I am taking AI and ML. Fortunately, I am allowed to spend some time doing this at work, but it is still taking my weekends. I think one class is ok, two is a bit too much but doable. I wouldn't do three.
A few coworkers and I signed up for the DB class and ran into the same problem. What we decided to do instead is just read through the material whenever we get a chance and not worry too much about the assignments (unless we get time). We don't necessarily get all the benefits, but at least we're able to take something good out of it.
That's what I've been doing as well. No "certificate of accomplishment" at the end of the class, but still the benefits of learning the material. Seems like a fair trade.
In my opinion, the real learning comes from doing the exercises. So if you are reading the material but not doing any quiz or exercise, you aren't learning as much. And if you aren't doing the exam, which has a time limit and you can't repeat until you get it perfect, you don't know how much you are or aren't learning.
Totally agree. I signed up for the DB course but could not keep up due to responsibilities of my full time job. However, I've kept the assignments and plan to do them later.
I signed up for AI and DB classes next to a 45 hour/week job, and dropped the AI course. I'm having trouble keeping up while maintaining a life, but it is possible, though!
Seriously, consider only one class if you have any trouble self-regulating.
Way to go Stanford. You guys just keep blowing our minds. After all the stuff i learned for free in just 4 weeks of the current classes, i am filled with gratitude! I highly recommend these classes..
This question might be tangent from the intended purpose of these awesome free classes but can anyone share if some certification or acknowledgement of some sort is received upon completion?
I think the recent AI, DB, and ML classes all planned to offer an acknowledgement of completion. This one, linked from the OP, mentioned something similar: http://www.pgm-class.org/
"Students will have access to lecture videos, lecture notes, receive regular feedback on progress, and receive answers to questions. When you successfully complete the class, you will also receive a statement of accomplishment."
ML-class does not necessarily apply to all other 10 courses offered by Stanford. I would not be surprised if they weren't given for the entrepreneurial classes offered.
One of the AI tutors tweeted about the possibility of offering a Master's degree for $2000. AFAIK, there hasn't been any more public discussion of this.
From the courses building up, it seems quite possible. The $2000 could be to cover limited personal support, and the sitting of a proper exam, with the bulk of the teaching coming from the online videos.
You will receive a statement from the Professor, but I believe they have to be careful about avoiding the perception that it is a certificate from Stanford university.
Yes, for the AI class they promised to give you a signed statement of accomplishment. IIRC, this statement will also tell you how well you did in relation to all other students (including Stanford students).
Applied Cryptography, no matter what Matthew Green may have to say about it, is a terrible book to learn cryptography from. I highly recommend you burn it and instead pick up a copy of Practical Cryptography (or Cryptography Engineering, which is the exact same book).
The mark of a good book on a security topic is, you can read it "upside down" and learn how to break things instead of build them.
Menezes is "Handbook Of Applied Cryptography", but this was actually a sore point with Schneier back in '98 (I feel like I listened to him complain about it in a bar at Usenix Security).
I disagree that it's the only comment with a rationale, but it's a fine example. Writing the same thing about the same subjects over and over again is tiresome, isn't it?
(Thank you for taking the time to pick one out though).
The AI class is ok, but the lack of programming assignment really impede learning. On the other hand the ML class is great, even the web site is better.
Heh, I had the same reaction after reading Neil Stephenson's Cryptonomicon. Working for an ISP and seeing how easy it is to sniff traffic only makes the paranoia worse, unfortunately.
That's really interesting and it seems, according to ml-class videos already online, that there are also subtitles available (it surely helps people like me that are not native english speaker)
other already pointed out that having assignments helps, I'd add that for some people (e.g. me) a _schedule_ is a great advantage.
If you are trying to learn something on your own timing it's easy to procrastinate. Maybe you could keep a faster pace but you just won't push yourself a bit.
It goes without saying that people with good self discipline can do without a schedule, but for some of us it helps.
I've had the opposite problem: Without a schedule, I'll binge on the material for the first few days, and then utterly burn out or fail to properly learn the information.
IMHO, there is an illusion that i understand stuff just by watching a video. i don't know how much of a noob i still am until exercises and midterms. The stanford model should be emulated...seriously.
I'd argue that the key motivator for hackers is the presence of interesting and creative course work that has a clear purpose. Schedules tend to arbitrarily restrict creativity and learning.
This means that courses should be designed around the concept of incrementally building something useful.
A "subject" would not be broken down into one specific topic area, but rather, would encompass a small project such as:
1. develop a library that performs rudimentary speech-to-text translation and then create a plugin for {selected open source software} that makes use of this library
2. build a geospatial database that stores OpenStreetMap data and then create a rudimentary website that {highlights areas that are not within a 3km diameter of a police station, etc}
3. analyse the implementation of cryptography in {selected open source software} against many avenues of attack, particularly side-channel attacks, entropy of random data sources, etc and prepare patches that solve the problems found
In the above examples, mostly-complete plugins or websites could be provided to students so they can focus on more important parts of each problem.
I feel this would work because each small project:
1. allows a great deal of creativity from students
2. encourages students to extend themselves throughout and _after_ the small course has been completed
3. doesn't force students to move on too early
4. demonstrates how to _apply_ knowledge they've gained
5. rewards students with something more than a number/grade as they progress
6. encourages entrepreneurship
n. ...
Many of these points are built into Kahn Academy by design. This is why the theory is well suited to Kahn Academy where you can map out skills and ensure you've ticked off dependencies before moving on. Their new grading system (a learning algorithm) helps students move on _when they're ready_, not at some arbitrarily defined point of time.
A course would therefore start by introducing the project, providing a demonstration of the expected outcomes and seeding some ideas for how the project could be extended and applied to unsolved problems.
Links to Kahn Academy would provide succinct theoretical background. Course material would apply that knowledge and demonstrate how it is used to do something useful to assist with completing the project.
Hopefully this form of education builds a desire to learn. Schedules and grading become irrelevant when students are having fun and making progress on interesting problems. Students will prioritise the project over other things in their life that they'd otherwise procrastinate with. The added benefit to society is increased levels of research and development, entrepreneurship and decreased education costs.
It's free and the assignments will be graded automatically by a computer. See http://www.ml-course.com for a currently running Stanford course example.
Also why are they not shifting away from CS classes (except for entrepreneurship). Things like signal processing and basic controllers can be taught with such a class. I would specially love the signal processing class since there are 3 classes (don't exactly remember the name) which are considered holy perfecta of signal processing.